MELTING - nearest-neighbor computation of

nucleic acid hybridation

Marine Dumousseau, William John Gowers
Nicolas Le Novère
[email protected]
February 2014

1
Contents
1 Synopsis 4

2 Description 4

3 Usage 4
3.1 Information about MELTING . . . . . . . . . . . . . . . . . . . . . . 5
3.2 Mandatory options . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.3 General options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.4 Set of thermodynamic parameters and methods (models) . . . . . . . 8

4 Algorithm 18
4.1 Thermodynamics of helix-coil transition of nucleic acid . . . . . . . . 18
4.1.1 Perfectly matching sequences . . . . . . . . . . . . . . . . . 19
4.1.2 Sequences composed of CNG repeats . . . . . . . . . . . . . 22
4.1.3 Single mismatch effect . . . . . . . . . . . . . . . . . . . . . 23
4.1.4 Tandem mismatches effect . . . . . . . . . . . . . . . . . . . 28
4.1.5 Internal loop effect . . . . . . . . . . . . . . . . . . . . . . . 31
4.1.6 GU wobble base pairs effect . . . . . . . . . . . . . . . . . . 33
4.1.7 Single dangling end effect . . . . . . . . . . . . . . . . . . . 34
4.1.8 Double dangling end effect . . . . . . . . . . . . . . . . . . . 37
4.1.9 Long dangling end effect (poly A) . . . . . . . . . . . . . . . 38
4.1.10 Single bulge loop effect . . . . . . . . . . . . . . . . . . . . 42
4.1.11 long bulge loop effect . . . . . . . . . . . . . . . . . . . . . 45
4.1.12 Inosine bases effect . . . . . . . . . . . . . . . . . . . . . . . 46
4.1.13 Azobenzenes effect . . . . . . . . . . . . . . . . . . . . . . . 49
4.1.14 2-Hydroxyadenine bases effect . . . . . . . . . . . . . . . . . 50
4.1.15 Single Locked nucleic acid effect . . . . . . . . . . . . . . . 52
4.1.16 Consecutive Locked nucleic acids effect . . . . . . . . . . . . 53
4.1.17 Consecutive Locked nucleic acids with a single mismatch effect 54
4.2 The melting temperature . . . . . . . . . . . . . . . . . . . . . . . . 55
4.3 Correction for the concentration of nucleic acid . . . . . . . . . . . . 55
4.4 Correction for the concentration of cations . . . . . . . . . . . . . . . 58
4.4.1 Sodium corrections . . . . . . . . . . . . . . . . . . . . . . . 58
4.4.2 Magnesium corrections . . . . . . . . . . . . . . . . . . . . . 61
4.4.3 Mixed Na Mg corrections . . . . . . . . . . . . . . . . . . . 62
4.5 Correction for the concentration of denaturing agents . . . . . . . . . 64
4.5.1 DMSO corrections, DMSO in % . . . . . . . . . . . . . . . . 64
4.5.2 formamide corrections . . . . . . . . . . . . . . . . . . . . . 65
4.6 Long sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

5 See Also 74

6 Copyright 75

7 Acknowledgements 75

2
8 Authors 75

9 History 75

3
1 Synopsis
The nearest-neighbor approach is based on the fact that the helix-coil transition works
as a zipper. After an initial attachment, the hybridisation propagates laterally. The
hybridization process depends on the adjacent nucleotides on each strand (the Crick’s
pairs). Two duplexes with the same base pairs could have different stabilities, and on
the contrary, two duplexes with different sequences but identical sets of Crick’s pairs
will have the same thermodynamics properties (see Sugimoto et al. 1994). See Wetmur
J.G (1991) and Santalucia (1998) for deep reviews on the nucleic acid hybridization and
on the different set of nearest-neighbor parameters.

2 Description
MELTING computes, for a nucleic acid duplex, the enthalpy and the entropy of the
helix-coil transition, and then its melting temperature. Four types of hybridisation
are possible: DNA/DNA, DNA/RNA, RNA/RNA and 2-O-Methyl RNA/RNA. The
program uses the method of nearest-neighbors. The set of thermodynamic parameters
can be easely changed, for instance following an experimental breakthrough. Melting
is a free program in both sense of the term. It comes with no cost and it is open-source.
In addition it is coded in Java (1.5) and can be compiled on any operating system.
If you use MELTING, please quote

Le Novère. MELTING, a free tool to compute the melting temperature of

nucleic acid duplex. Bioinformatics, 17: 1226-1227.

Dumousseau M, Rodriguez N, Juty N, Le Novère N. MELTING, a flex-

ible platform to predict the melting temperatures of nucleic acids. BMC
Bioinformatics, 16;13:101, PMID: 22591039.

3 Usage
The options are treated sequentially. If there is a conflict between the value of two
options, the latter normally erases the former.

BE AWARE : The option syntax of MELTING 5 is different from the one of MELT-
ING 4. There is a space between the option name and the option value. New option
names are available in MELTING 5 to change the default thermodynamic models and
default corrections.
There is no interactive mode in MELTING 5 therefore the option ’-q’ doesn’t exist
anymore.
You can use the MELTING 4 option syntax, but it doesn’t allow the user to change
some of the thermodynamic models and corrections. In addition to that, the user can’t
enter a formamide or DMSO See the README file to choose the adapted executable.
The MELTING 4 option name ’-x’ is equivalent to the MELTING 5 option name ’-am’.
There is no input file option in MELTING 5 (option ’-I’) but you can use this option

4
for the compatible executable of MELTING 5.
The MELTING 4 option names ’-N’, ’-t’, ’-k’, ’G’ are replaced by the single option
’-E’ in MELTING 5.
The MELTING 4 option names ’-A’, ’-D’, ’-M’ are respectively equivalent to ’-nn’,
’-sinDE’, ’-sinMM’ in MELTING 5.
The file names to write with the precedent option are replaced by thermodynamic
model names (see below).
The MELTING 4 option name ’-K’ is replaced by ’-ion’ in MELTING 5.

3.1 Information about MELTING

-h
Displays a short help and quit.
-L
Prints the legal informations and quit.
-V
Displays the version number and quit.
-p
Return the directory supposed to contain the sets of calorimetric parameters and
quit. If the environment variable NN_PATH is set, it is returned. Otherwise, the
value defined by default during the compilation is returned.

3.2 Mandatory options

-S sequence
Sequence of one strand of the nucleic acid duplex, entered 5’ to 3’. Important:
Uridine and thymidine are not considered as identical. The bases can be upper
or lowercase.

-C complementary_sequence
Enters the complementary sequence, from 3’ to 5’. This option is mandatory if
there are mismatches, inosine(s) or hydroxyadenine(s) between the two strands.
If it is not used, the program will compute it as the complement of the sequence
entered with the option -S. In case of self complementary sequences, The pro-
gram can automatically detect the symmetry and deduce the complementary even
though there is (are) dangling end(s) and it is not necessary to write the comple-
mentary sequence with the option -C. Uridine and thymidine are not considered
as identical. The bases can be upper or lowercase.
-E ion1_name=x.xxe-xx:ion2_name=x.xxe-xx:agent1_name=x.xxe-xx...
Enters the different ion (Na, Mg, Tris, K) or agent (dNTP, DMSO, formamide)
concentrations. The effect of ions and denaturing agents on thermodynamic sta-
bility of nucleic acid duplexes is complex, and the correcting functions are at

5
 best rough approximations. All the concentrations must be positive numeric val-
ues and in M. There are some exceptions for the DMSO concentrations (in %)
and the formamide concentrations (in % or M depending on the used correction
method). Be aware, the [Tris+ ] is about half of the total tris buffer concentration.
At least one cation concentration is mandatory, the other agents are optional. See
the documentation for the concentration limits. It depends on the used correc-
tion.
-P x.xxe-xx
Concentration of the nucleic acid strand in excess. It must be a strict positive
numeric value and it is mandatory. The oligomer concentration is in mol/L.

-H hybridisation_type
Specifies the hybridisation type. Moreover this parameter determines the nature
of the sequences entered by the user. Possible values are :
dnadna : DNA sequence (option -S) and DNA complementary sequence (option
-C)
rnarna : RNA sequence (option -S) and RNA complementary sequence (option
-C)
dnarna : DNA sequence (option -S) and RNA complementary sequence (option
-C)
rnadna : RNA sequence (option -S) and DNA complementary sequence (option
-C)
mrnarna : 2-o-methyl RNA sequence (option -S) and RNA complementary sequence
(option -C)
mrnarna : RNA sequence (option -S) and 2-o-methyl RNA complementary sequence
(option -C)

This option is mandatory to select the default equations and methods to use.

3.3 General options

-T xxx
Size threshold before approximative computation. The nearest-neighbour ap-
proach will be used by default if the length of the sequence is inferior to this
threshold, otherwise it is the approximative approach which will be used by de-
fault.

-v
Activates the verbose mode, issuing a lot more information about the current run
(try it once to see if you can get something interesting).
-nnpath folder_pathway
Change the default pathway (Data) where to find the default calorimetric tables
(thermodynamic parameters). The program will look for the file in a directory

6
 specified during the installation. However, if an environment variable NN_PATH
is defined, melting will search in this one first.
-O output_file
The output is directed to this file instead of the standard output. The name of the
file must be specified.
-self
To precise that the sequence entered with the option -S is self complementary.
No complementary sequence is mandatory. The program automatically can de-
tect a self complementary sequence for perfect matching sequences or sequences
with dangling ends. In these cases, the option -self is not necessary. Otherwise
we need to precise that the sequences are self complementary with this option.
examples:

Situation 1 : The sequence ATCGCGAT is self

complementary.

The option -self is not necessary because the program can automatically detect
it.

Situation 2 : The sequence -TCGCGAT is self

complementary with a single
dangling end.

The option -self is not necessary because the program can automatically detect
it.

Situation 3 : If the sequence ATCCCGAT is self

complementary with a single mismatch
(C/C)

The option -self is necessary to precise the self complementarity because the
program can’t detect it.
-F factor
This is the correction factor used to modulate the effect of the nucleic acid con-
centration in the computation of the melting temperature. See section ALGO-
RITHM for details. If the sequences are automatically recognized as self com-
plementary sequences or if the option -self is used, the factor correction is auto-
matically 1. Otherwise F is 4 if the both strands are present in equivalent amount
and 1 if one strand is in excess. The default factor value is 4.

7
 3.4 Set of thermodynamic parameters and methods (models)
By default, the approximative mode is used for oligonucleotides longer than 60 bases
(the default threshold value), otherwise the nearest neighbor model is used.
-am method_name
Forces to use a specific approximative formula, based on G+C content. You can
use one of the following :
DNA DUPLEXES
ahs01 (from von Ahsen et al. 2001)
che93 (from Marmur 1962„ Chester and al. 1993)
che93corr (from von Ahsen et al. 2001, Marmur 1962, Chester and al. 1993)
schdot (Marmur-Schildkraut-Doty formula)
owe69 (from Owen et al. 1969)
san98 (from Allawi and Santalucia. 1998)
wetdna91 (from Wetmur 1991) (by default)
RNA DUPLEXES
wetrna91 (from Wetmur 1991) (by default)
DNA/RNA DUPLEXES
wetdnarna91 (from Wetmur 1991) (by default)
If there is no formula name after the option -am, we will compute the melting
temperature with the default approximative formula. This option has to be used
with caution. Note that such a calcul is increasingly incorrect when the length
of the duplex decreases. Moreover, it does not take into account nucleic acid
concentration, which is a strong mistake. examples :

command line 1 : "-am"

if you want to force the approximative approach with the default formula.

command line 2 : "-am ahs01"

if you want to use the approximative formula from Ahsen et al. 2001.
-nn method_name
Forces to use a specific nearest neighbor model. You can use one of the following
:
DNA DUPLEXES

8
all97 (from Allawi and Santalucia 1997) (by default)
bre86 (from Breslauer et al. 1986)
san04 (from Hicks and Santalucia 2004)
san96 (from Santalucia et al. 1996)
sug96 (from Sugimoto et al 1996)
tan04 (from Tanaka et al. 2004)
RNA DUPLEXES
fre86 (from Freier al. 1986)
xia98 (from Xia et al. 1998) (by default)
DNA/RNA DUPLEXES
sug95 (from Sugimoto et al. 1995) (by default)
M RNA/RNA DUPLEXES

tur06 (from Kierzek et al. 2006) (by default)

If there is no formula name after the option -nn, we will compute the melting
temperature with the default nearest neighbor model. Each nearest neighbor
model uses a specific xml file containing the thermodynamic values. If you want
to use another file, write the file name or the file pathway preceded by ’:’ (-nn
[optionalname:optionalfile]). examples:

Command line 1 : "-nn"

if you want to force the nearest neighbor computation with the default model.

Command line 2 : "-nn tan04"

if you want to use the nearest neighbor model from Tanaka et al. 2004 with the
thermodynamic parameters in the default xml file.

Command line 3 : "-nn tan04:fileName"

if you want to use the nearest neighbor model from Tanaka et al. 2004 with the
thermodynamic parameters in the file fileName.

9
 Command line 4 : "-nn :fileName"

if you want to use the default nearest neighbor model with the thermodynamic
parameters in the file fileName.
-sinMM method_name
Forces to use a specific nearest neighbor model to compute the contribution of
single mismatch to the thermodynamic of helix-coil transition. You can use one
of the following :
DNA DUPLEXES
allsanpey (from Allawi, Santalucia and Peyret 1997, 1998 and 1999) (by default)

DNA/RNA DUPLEXES
wat10 (from Watkins et al. 2011) (by default)
RNA DUPLEXES
tur06 (from Lu et al. 2006)
zno07 (from Davis et al. 2007) (by default)
zno08 (from Davis et al. 2008)
To change the file containing the thermodynamic parameters for single mismatch
computation, the same syntax as the one for the -nn option is used. Single mis-
matches are not taken into account by the approximative mode.
-GU method_name
Forces to use a specific nearest neighbor model to compute the contribution of
GU base pairs to the thermodynamic of helix-coil transition. You can use one of
the following :
RNA DUPLEXES
tur99 (from Mathews et al. 1999)
ser12 (from Serra et al. 2012) (by default)
To change the file containing the thermodynamic parameters for GU base pair
computation, the same syntax as the one for the -nn option is used. GU base
pairs are not taken into account by the approximative mode.
-tanMM method_name
Forces to use a specific nearest neighbor model to compute the contribution of
tandem mismatches to the thermodynamic of helix-coil transition. You can use
one of the following :
DNA DUPLEXES

10
allsanpey (from Allawi, Santalucia and Peyret 1997, 1998 and 1999) (by default)
RNA DUPLEXES
tur99 (from Mathews et al. 1999) (by default)
To change the file containing the thermodynamic parameters for tandem mis-
match computation, the same syntax as the one for the -nn option is used. Tan-
dem mismatches are not taken into account by the approximative mode. Note
that not all the mismatched Crick’s pairs have been investigated.
-intLP method_name
Forces to use a specific nearest neighbor model to compute the contribution of
internal loop to the thermodynamic of helix-coil transition. You can use one of
the following :
DNA DUPLEXES
san04 (from Hicks and Santalucia 2004) (by default)

RNA DUPLEXES
tur06 (from Lu et al. 2006) (by default)
zno07 (from Badhwar et al. 2007, only for 1x2 loop)
To change the file containing the thermodynamic parameters for internal loop
computation, the same syntax as the one for the -nn option is used. Internal
loops are not taken into account by the approximative mode.
-sinDE method_name
Forces to use a specific nearest neighbor model to compute the contribution of
single dangling end to the thermodynamic of helix-coil transition. You can use
one of the following :
DNA DUPLEXES
bom00 (from Bommarito et al. 2000) (by default)
sugdna02 (from Ohmichi et al. 2002, only for polyA dangling ends)

RNA DUPLEXES
sugrna02 (from Ohmichi et al. 2002, only for polyA dangling ends)
ser08 (from Miller et al. 2008) (by default)
To change the file containing the thermodynamic parameters for single dangling
end computation, the same syntax as the one for the -nn option is used. Single
dangling ends are not taken into account by the approximative mode.

11
-secDE method_name
Forces to use a specific nearest neighbor model to compute the contribution of
double dangling end to the thermodynamic of helix-coil transition. You can use
one of the following :
DNA DUPLEXES

sugdna02 (from Ohmichi et al. 2002, only for polyA dangling ends) (by default)
RNA DUPLEXES
sugrna02 (from Ohmichi et al. 2002, only for polyA dangling ends)
ser05 (from O’toole et al. 2005)
ser06 (from O’toole et al. 2006) (by default)
To change the file containing the thermodynamic parameters for double dangling
end computation, the same syntax as the one for the -nn option is used. Double
dangling ends are not taken into account by the approximative mode.

-longDE method_name
Forces to use a specific nearest neighbor model to compute the contribution of
long dangling end to the thermodynamic of helix-coil transition. You can use
one of the following :
DNA DUPLEXES

sugdna02 (from Ohmichi et al. 2002, only for polyA dangling ends) (by default)
RNA DUPLEXES
sugrna02 (from Ohmichi et al. 2002, only for polyA dangling ends)
To change the file containing the thermodynamic parameters for long dangling
end computation, the same syntax as the one for the -nn option is used. Long
dangling ends are not taken into account by the approximative mode.
-sinBU method_name
Forces to use a specific nearest neighbor model to compute the contribution of
single bulge loop to the thermodynamic of helix-coil transition. You can use one
of the following :
DNA DUPLEXES
san04 (from Santalucia 2004)
tan04 (from Tanaka et al. 2004) (by default)

RNA DUPLEXES
ser07 (from Blose et al. 2007)
tur06 (from Lu et al. 1999 and 2006) (by default)

12
 To change the file containing the thermodynamic parameters for single bulge
loop computation, the same syntax as the one for the -nn option is used. Single
bulge loops are not taken into account by the approximative mode.
-lonBU method_name
Forces to use a specific nearest neighbor model to compute the contribution of
long bulge loop to the thermodynamic of helix-coil transition. You can use one
of the following :
DNA DUPLEXES
san04 (from Hicks and Santalucia 2004) (by default)

RNA DUPLEXES
tur06 (from Mathews et al. 1999 and Lu et al 2006) (by default)
To change the file containing the thermodynamic parameters for long bulge loop
computation, the same syntax as the one for the -nn option is used. Long bulge
loops are not taken into account by the approximative mode.

-CNG method_name
Forces to use a specific nearest neighbor model to compute the contribution of
CNG repeats to the thermodynamic of helix-coil transition. N represents a single
mismatch of type N/N. You can use one of the following :
RNA DUPLEXES

bro05 (from Magdalena et al. 2005) (by default)

To change the file containing the thermodynamic parameters for CNG repeats
computation, the same syntax as the one for the -nn option is used. CNG repeats
are not taken into account by the approximative mode. Be aware : Melting can
compute the contribution of CNG repeats to the thermodynamic of helix-coil
transition for only 2 to 7 CNG repeats.
-ino method_name
Forces to use a specific nearest neighbor model to compute the contribution of
inosine bases (I) to the thermodynamic of helix-coil transition. You can use one
of the following :
DNA DUPLEXES
san05 (from Watkins and Santalucia 2005) (by default)
RNA DUPLEXES

zno07 (from Wright et al. 2007, only IU base pairs) (by default)
To change the file containing the thermodynamic parameters for inosine bases
computation, the same syntax as the one for the -nn option is used. Inosine
bases (I) are not taken into account by the approximative mode.

13
-ha method_name
Forces to use a specific nearest neighbor model to compute the contribution of
hydroxyadenine bases (A*) to the thermodynamic of helix-coil transition. You
can use one of the following :
DNA DUPLEXES

sug01 (from Kawakami et al. 2001) (by default)

To change the file containing the thermodynamic parameters for hydroxyadenine
bases computation, the same syntax as the one for the -nn option is used. Hy-
droxyadenine bases (A*) are not taken into account by the approximative mode.

-azo method_name
Forces to use a specific nearest neighbor model to compute the contribution of
azobenzenes (X_T for trans azobenzenes and X_C for cis azobenzenes) to the
thermodynamic of helix-coil transition. You can use one of the following :
DNA DUPLEXES

asa05 (from Asanuma et al. 2005)(by default)

To change the file containing the thermodynamic parameters for azobenzene
computation, the same syntax as the one for the -nn option is used. Azoben-
zenes (X_T for trans azobenzenes and X_C for cis azobenzenes) are not taken
into account by the approximative mode.

-lck method_name
Forces to use a specific nearest neighbor model to compute the contribution of
single locked nucleic acids (AL, GL, TL and CL) to the thermodynamic of helix-
coil transition. You can use one of the following :
DNA DUPLEXES

mct04 (from McTigue et al. 2004)

owc11 (from Owczarzy et al. 2011) (by default)
To change the file containing the thermodynamic parameters for single locked
nucleic acids computation, the same syntax as the one for the -nn option is used.
Locked nucleic acids (AL, GL, TL and CL) are not taken into account by the
approximative mode.
-tanLck method_name
Forces to use a specific nearest neighbor model to compute the contribution of
consecutive locked nucleic acids (AL, GL, TL and CL) to the thermodynamic of
helix-coil transition. You can use one of the following :
DNA DUPLEXES
owc11 (from Owczarzy et al. 2011) (by default)

14
 To change the file containing the thermodynamic parameters for consecutive
locked nucleic acids computation, the same syntax as the one for the -nn option
is used. Locked nucleic acids (AL, GL, TL and CL) are not taken into account
by the approximative mode.
-sinMMLck method_name
Forces to use a specific nearest neighbor model to compute the contribution of
single mismatch in consecutive locked nucleic acids (AL, GL, TL and CL) to the
thermodynamic of helix-coil transition. You can use one of the following :
DNA DUPLEXES
owc11 (from Owczarzy et al. 2011) (by default)

To change the file containing the thermodynamic parameters for single mismatch
in consecutive locked nucleic acids computation, the same syntax as the one for
the -nn option is used. Locked nucleic acids (AL, GL, TL and CL) are not taken
into account by the approximative mode.

-ion method_name
Forces to use a specific ion correction. You can use one of the following correc-
tions :
Sodium corrections
DNA DUPLEXES

ahs01 (from von Ahsen et al. 2001)

kam71 (from Frank-Kamenetskii 2001)
owc1904 (equation 19 from Owczarzy et al. 2004)
owc2004 (equation 20 from Owczarzy et al. 2004)
owc2104 (equation 21 from Owczarzy et al. 2004)
owc2204 (equation 21 from Owczarzy et al. 2004) (by default)
san96 (from Santalucia et al. 1996)
san04 (from Santalucia et al. 1998, 2004)
schlif (from Schildkraut and Lifson 1965)
tanna06 (from Tan et al. 2006)
wetdna91 (from wetmur 1991)
RNA DUPLEXES OR M RNA/RNA DUPLEXES
tanna07 (from Tan et al. 2007) (by default)
wetrna91 (from wetmur 1991)
DNA/RNA DUPLEXES
wetdnarna91 (from wetmur 1991)

15
 Magnesium corrections
DNA DUPLEXES
owcmg08 (from Owczarzy et al. 2008) (by default)
tanmg06 (from Tan et al. 2006)

RNA DUPLEXES OR M RNA/RNA DUPLEXES

tanmg07 (from Tan et al. 2007) (by default)
Mixed Na Mg corrections
DNA DUPLEXES

owcmix08 (from Owczarzy et al. 2008) (by default)

tanmix07 (from Tan et al. 2007)
RNA DUPLEXES OR M RNA/RNA DUPLEXES
tanmix07 (from Tan et al. 2007) (by default)

The effect of ions on thermodynamic stability of nucleic acid duplexes is com-

plex, and the correcting functions are at best rough approximations. By default,
the program use the algorithm from Owczarzy et al 2008 : ratio = [Mg0.5 ] and
monovalent = Na + Tris + K
if monovalent = 0, a magnesium correction is used.
if ratio < 0.22, a sodium correction is used.
if 0.22 <= ratio < 6, a mixed Na Mg correction is used.
if ratio >= 6, a magnesium correction is used.
example :

Command line : "-ion owcmg08"

if you want to force the use of the magnesium correction from Owczarzy et al
2008. This correction will be used independently of the cations present in the
solution.
-naeq method_name
Forces to use a specific ion correction which gives a sodium equivalent concen-
tration if other cations are present. You can use one of the following :
DNA DUPLEXES
ahs01 (from von Ahsen et al 2001) (by default)
mit96 (from Mitsuhashi et al. 1996)

16
 pey00 (from Peyret 2000)
For the other types of hybridization, the DNA default correction is used but there
is no guaranty of accuracy. If there are other cations when an approximative ap-
proach is used, a sodium equivalence is automatically computed. The correcting
functions are at best rough approximations. example :

Command line 1 : "-naeq ahs01"

if you want to force the use of the sodium equivalence from Ahsen et al 2001.
This sodium equivalence will be used in case of approximative approach. In
case of nearest neighbor approach, the sodium equivalence will be used only if a
sodium correction is selected by the user.

Command line 2 : "-naeq ahs01 -ion san04"

it means that the sodium equivalence computed by the method ahs01 (from Ah-
sen et al 2001) will be combined with the sodium correction san04 (from San-
talucia 2004).
-DMSO method_name
Forces to use a specific DMSO correction (DMSO is always in %). You can use
one of the following :
DNA DUPLEXES
ahs01 (from von Ahsen et al 2001) (by default)
mus81 (from Musielski et al. 1981)
cul76 (from Cullen et al. 1976)
esc80 (from Escara et al. 1980)
For the other types of hybridization, the DNA default correction is used but there
is no guaranty of accuracy. If there are DMSO when an approximative approach
is used, a DMSO correction is automatically computed. The correcting functions
are at best rough approximations. example :

Command line : "-DMSO ahs01"

if you want to force the use of the DMSO correction from Ahsen et al 2001. This
DMSO correction will be used if there is DMSO present in the solutions in case
of nearest neighbor approach and approximative approach.

17
-for method_name
Forces to use a specific formamide correction. You can use one of the following
:
DNA DUPLEXES
bla96 (from Blake 1996) with formamide concentration in mol/L (by default)
lincorr (linear correction) with a % of formamide volume
For the other types of hybridization, the DNA default correction is used but there
is no guaranty of accuracy. If there are formamide when an approximative ap-
proach is used, a formamide correction is automatically computed. The correct-
ing functions are at best rough approximations. example :

Command line : "-for lincorr"

if you want to force the use of the linear formamide correction. This formamide
correction will be used if there is formamide present in the solutions in case of
nearest neighbor approach and approximative approach.

4 Algorithm
4.1 Thermodynamics of helix-coil transition of nucleic acid
The nearest-neighbor approach is based on the fact that the helix-coil transition works
as a zipper. After an initial attachment, the hybridisation propagates laterally. This
program first computes the hybridisation enthalpy and entropy for each structure in the
duplex. (see later for the different possible structures recognized by Melting). If the
sequences are self complementary, a symmetry correction will be added to the initiation
energy.
∆H = δ hinitiation + ∑ δ hstructure
∆S = δ sinitiation + ∑ δ sstructure
Example :
Sequence with a single mismatch
ATCGGCTA
TAGACGAT
∆H = δ hinitiation + δ hstructure1 + δ hstructure2 + δ hstructure3
∆S = δ sinitiation + δ sstructure1 + δ sstructure2 + δ sstructure3
where :
structure1 = perfectly matching sequences ATC/TAG
structure2 = single mismatch G/A
structure 3 = perfectly matching sequences GCTA/CGAT

18
4.1.1 Perfectly matching sequences
The hybridization process depends on the adjacent nucleotides on each strand (the
Crick’s pairs). Two duplexes with the same base pairs could have different stabilities,
and on the contrary, two duplexes with different sequences but identical sets of Crick’s
pairs will have the same thermodynamics properties. This program first computes the
hybridisation enthalpy and entropy from the elementary parameters of each Crick’s
pair.
∆hperfectly−matching = ∑ δ hCrick0 spair
∆sperfectly−matching = ∑ δ sCrick0 spair
The initiation computation is not the same for each following model.

model limits Article

all97 DNA Allawi and SantaLucia (1997)
Biochemistry 36 : 10581-10594
bre86 DNA Breslauer et al. (1986)
Proc Natl Acad Sci USA 83 : 3746-3750
san04 DNA Santalucia and Hicks (2004)
Annu. Rev. Biophys. Biomol. Struct 33 : 415-440
san96 DNA SantaLucia et al.(1996)
Biochemistry 35 : 3555-3562
sug96 DNA Sugimoto et al. (1996)
Nuc Acids Res 24 : 4501-4505
tan04 DNA Tanaka et al (2004)
Biochemistry 43 : 7143-7150
fre86 RNA Freier et al (1986)
Proc Natl Acad Sci USA 83: 9373-9377
xia98 RNA Xia et al (1998)
Biochemistry 37: 14719-14735
sug95 DNA/RNA SantaLucia et al.(1996)
Biochemistry 35 : 3555-3562
tur06 mRNA/RNA Kierzeck et al (2006)
A sodium Nucleic acids research 34: 3609-3614
correction (san04)
is automatically
applied to
the computed
entropy to
convert the
entropy (Na = 0.1M)
into the
entropy (Na=1M)

Example :

19
 AGCGA AG GC CG GA
         
∆H = ∆H + ∆H + ∆H + ∆H
TCGCT TC CG GC CT
(The same computation is performed for ∆S)

Figure 1: Comparison of experimental and computed Tm for various sets of DNA

nearest-neighbor parameters. [Na+ ] = 1 M, [nucleic acid] = 4 · 10−4 M

Figure 2: Comparison of experimental and computed Tm for various sets of RNA

nearest-neighbor parameters. [Na+ ] = 1 M, [nucleic acid] = 2 · 10−4 M

20
Figure 3: Comparison of experimental and computed Tm for various sets of DNA/RNA
nearest-neighbor parameters. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

Figure 4: Comparison of experimental and computed Tm for various sets of 2-O-

methyl RNA nearest-neighbor parameters. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

21
4.1.2 Sequences composed of CNG repeats
If the sequence (sens 5’3’) is a sequence of type G(CNG)xC where x is the number of
CNG repeats in the sequence and N a unique nucleic acid which will get bound to itself,
we can use specific experimental parameters to compute the enthalpy and entropy of
the duplex formation. These parameters can be used only for sequences composed
from 2 to 7 CNG repeats and the initiation is already included.

∆H = ∆hsequence−of−type−G(CNG)xC
For further information, see the referenced article.

model limits Article

bro05 RNA Magdalena et al (2005)
Self complementary sequences Biochemistry 44: 10873-10882
2 to 7 CNG repeats

Example :

GCAGCAGCAGC
CGACGACGACG

GCAGCAGCAGC
 
∆H = ∆H(3-CAG-repeats)
CGACGACGACG
(The same computation is performed for ∆S)

22
Figure 5: Comparison of experimental and computed Tm for various sets of RNA
sequences composed of CNG repeats. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

Be aware : The results for sequences composed of 4 or 5 CCG repeats is not reli-
able. (the figure shows two values far from the expected temperature). This might be
due to a majority of hairpin loop formation. See the article above for further informa-
tions.

4.1.3 Single mismatch effect

The single mismatches are taken into account but the two first and positions cannot be
mismatched. in such a case, the result is unpredictable, and all cases are possible. for
instance (see Allawi and SanLucia 1997), the duplex
A T
GTGAGCTCAT
TACTCGAGTG
T A
is more stable than

AGTGAGCTCATT
TTACTCGAGTGA

23
 For DNA duplexes, this program computes the hybridisation enthalpy and entropy
from the elementary parameters of each Crick’s pair containing the single mismatch.

∆hsingle−mismatch = ∑ δ hCrick0 s−pair−containing−the−mismatch

Example :

ATC AT TC
     
∆H = ∆H + ∆H
TCG TC CG

(The same computation is performed for ∆S)

Figure 6: Comparison of experimental and computed Tm for various sets of DNA

sequences containing one single mismatch. [Na+ ] = 1 M, [nucleic acid] = 4 · 10−4 M

For DNA/RNA duplexes, the same model is used, taking parameters from Watkins
et al. (2011). The only mismatches permitted are dA/rA, dT/rU, dC/rC and dG/rG.

Example :

CAT CA AT
     
∆H = ∆H + ∆H
GAU GA AU

(The same computation is performed for ∆S)

24
Figure 7: Comparison of experimental and computed Tm for various sets of DNA/RNA
sequences containing one single mismatch. [Na+ ] = 1 M, [nucleic acid] = 4 · 10−4 M

For RNA duplexes, the different models to computes the thermodynamic contribu-
tion of single mismatch to the helix coil stability are more complex.

Model from Amber R. Davis and Brent M Znosco, 2007-2008

∆h(single-mismatch) = δ hmismatch−nucleotides + δ hmismatch−NN−interaction + δ hAU/GU

Where :
δ hmismatch−nucleotides accounts for the identity of the single mismatch nucleotides.
δ hmismatch−NNinteraction accounts for the interaction between the mismatch nucleotides
and the nearest neighbors. (R purine, Y pyrimidine)
δ hAU/GU accounts for AU or GU nearest neighbors.
Example :

AUC U RYY
     
∆H = ∆H + 1 × ∆H AU + ∆H
UUG U YYR

(The same computation is performed for ∆S)

Model from Zhi Johm Lu, Douglas H. Turner and David H. Mathews, 2006

∆h(single-mismatch) = δ hinitiation−loop−of−2 + δ hper−AU/GU + δ hGG + δ hRU/YU

25
 Where :
δ hinitiation−loop−of−2 accounts for the initiation of a single non canonical pair.
δ hGG accounts for a GG single mismatch.
δ hRU/YU accounts for a 5’RU/3’YU stack with R a purine and Y a pyrimidine.
δ hper−AU/GU accounts for AU or GU nearest neighbors.
Example :

AUC RU
   
∆H = ∆Hinitiation-loop-of-2 + 1 × ∆H per-AU + ∆H
UUG YU

(The same computation is performed for ∆S)

For further information, see the referenced articles.

model limits Article

allsanpey DNA Allawi and SantaLucia (1997)
Biochemistry 36: 10581-10594
Allawi and SantaLucia (1998)
Biochemistry 37: 2170-2179
Allawi and SantaLucia (1998)
Nuc Acids Res 26: 2694-2701
Allawi and SantaLucia (1998)
Biochemistry 37: 9435-9444
Peyret et al. (1999)
Biochemistry 38: 3468-3477
wat11 DNA/RNA Watkins et al. (2011)
tur06 RNA Lu et al (2006)
Nucleic Acids Research 34: 4912-4924
zno07 RNA Davis and Znosko (2007)
Biochemistry 46: 13425-13436
zno08 RNA Davis and Znosko (2008)
at least Biochemistry 47: 10178-10187
one adjacent
GU base pair

26
Figure 8: Comparison of experimental and computed Tm for various sets of RNA
sequences containing one single mismatch. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

27
4.1.4 Tandem mismatches effect
The tandem mismatches (two adjacent mismatches) are taken into account but the two
first and positions cannot be mismatched. Moreover the thermodynamic parameters
are still not available for every possible cases. In such a case, the program, unable to
compute any relevant result, will quit with a warning.
For DNA duplexes, this program computes the hybridisation enthalpy and entropy
from the elementary parameters of each Crick’s pair containing the mismatch(es).

∆htandem−mismatch = δ hCrick0 s−pair−containing−tandem−mismatch

+ ∑ δ hCrick0 s−pair−containing−single−mismatch

Example :

ATGC AT TG GC
       
∆H = ∆H + ∆H + ∆H
TCAG TC CA AG

(The same computation is performed for ∆S)

Figure 9: Comparison of experimental and computed Tm for various sets of DNA

sequences containing one tandem mismatch. [Na+ ] = 1 M, [nucleic acid] = 4 · 10−4 M

For RNA duplexes, the different models to computes the thermodynamic contribu-
tion of tandem mismatch to the helix coil stability are more complex.

28
Symmetric tandem mismatches : Model from Zhi Johm Lu, Douglas H. Turner
and David H. Mathews, 2006

∆h(tandem-mismatch) = δ htandem−mismatch+closing−base−pairs

Where :
δ hmismatch−nucleotides accounts for the identity of the double mismatch nucleotides
and the identity of the base pairs adjacent to the tandem mismatches.
Example :

G AC C AC
   
∆H = ∆H − ad jacent − to − GC
C CA G CA

(The same computation is performed for ∆S)

Asymmetric tandem mismatches : Model from Zhi Johm Lu, Douglas H. Turner
and David H. Mathews, 2006

δ hsymmetric−duplex−2 )
∆h(tandem-mismatch) = (δ hsymmetric−duplex−1 +
2
+ δ hGG + δ hp

Where :
δ hsymmetric−duplex−1 accounts for the enthalpy of a symmetric tandem mismatch
composed of the first closing base pair and the first mismatch nucleotides.
δ hsymmetric−duplex−2 accounts for the enthalpy of a symmetric tandem mismatch
composed of the second closing base pair and the second mismatch nucleotides.
δ hGG accounts for a GG pair adjacent to a AA pair or any non canonical pair
containing a pyrimidine.
δ hp accounts for an AG or GA pairs adjacent to a UC, CC or CU pair and a UU
pair adjacent to an AA pair .
Example :

A GC C A GA U G UC C
     
∆H = (∆H + ∆H + ∆H GA − ad jacent − to − CU
U AU G U AG A C CU G

(The same computation is performed for ∆S)

For further information, see the referenced articles.

29
 model limits Article
allsanpey DNA Allawi and SantaLucia (1997)
only GT Biochemistry 36: 10581-10594
mismatches Allawi and SantaLucia (1998)
and TA/TG Biochemistry 37: 2170-2179
mismatches Allawi and SantaLucia (1998)
Nuc Acids Res 26: 2694-2701
Allawi and SantaLucia (1998)
Biochemistry 37: 9435-9444
Peyret et al. (1999)
Biochemistry 38: 3468-3477
tur99 RNA Mathiews et al (1999)
no adjacent J.Mol.Biol. 288: 911-940
GU or UG base
pairs

Figure 10: Comparison of experimental and computed Tm for various sets of RNA
sequences containing one tandem mismatch. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

30
4.1.5 Internal loop effect
The internal loops (more than two adjacent mismatches) are taken into account but the
two first and positions cannot be mismatched. Moreover the thermodynamic parame-
ters are still not available for every possible cases. In such a case, the program, unable
to compute any relevant result, will quit with a warning. Moreover, the thermodynam-
ics of the nucleic acids within the internal loop are salt independent and no salt correc-
tion will be applied to it. However, the thermodynamics of the terminal mismatches
are salt dependent and a salt correction will be applied to them. The thermodynamic
model for DNA and RNA duplexes are similar.

DNA duplexes :Model from John Santalucia, Jr. and Donald Hicks, 2004

∆h(internal-loop(n)) = δ hasymmetry + δ hleft−terminal−mismatch

+ δ hright−terminal−mismatch
∆s(internal-loop(n)) = δ sloop(n) + δ sasymmetry + δ sleft−terminal−mismatch
+ δ sright−terminal−mismatch

Where :
δ hinternal−loop(n) accounts for the internal loop of n nucleotides.
δ hasymmetry accounts for the internal loop asymmetry (when the number of nucleic
acid within the internal loop is higher in one of the strand).
δ hleft−terminal−mismatch accounts for the identity of the first mismatch nucleotides of
the loop.
δ hright−terminal−mismatch accounts for the identity of the last mismatch nucleotides of
the loop.
Example : Symmetric internal loop

G ACCG C GA GC
     
∆H = ∆H + ∆H
C CATA G CC AG
G ACCG C GA GC
     
∆S = ∆Sloop of 8 + ∆S + ∆S
C CATA G CC AG

RNA duplexes :Model from Zhi Johm Lu, Douglas H. Turner and David H. Math-
ews, 2006

∆h(internal-loop(n)) = δ hinitiation−loop(n) + δ hper−AU/GU + (n1 − n2)δ hasymmetry

+ δ hfirst−non−canonical−pairs

Where :
δ hinitiation−loop(n) accounts for the internal loop of n nucleotides.

31
 δ hasymmetry accounts for the internal loop asymmetry (when the number of there
is an unequal numbers of nucleotides on each side) with n1 and n2 the number of
nucleotides on each strand..
δ hperA U/GU accounts for each AU or GU base pair adjacent to the internal loop.
δ hfirst−non−canonical−pairs accounts for each sequence specific first mismatch (bonus).
It is not applied to loops of the form 1 x (n-1) with n > 2.
Example : asymmetric internal loop

A ACCG C
 
∆H = ∆Hloop initiation(7) + 1 × ∆H per-AU + (4 − 3)∆Hasymmetry
U C-UA G

(The same computation is performed for ∆S)

For further information, see the referenced articles.

model limits Article

san04 DNA Santalucia and Hicks (2004)
missing asymmetry Annu. Rev. Biophys. Biomol. Struct 33 : 415-440
penalty,
not tested
with experimental
results
tur06 RNA Lu et al (2006)
not tested Nucleic Acids Research 34: 4912-4924
with experimental
results

32
Figure 11: Comparison of experimental and computed Tm for various sets of RNA
sequences containing one 1x2 internal loop. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

4.1.6 GU wobble base pairs effect

The wobble GU base pairs are taken into account. This pairing is a non-Watson-Crick
base pairing between two nucleotides in RNA molecules, but the thermodynamic sta-
bility of a wobble base pair is comparable to that of a Watson-Crick base pair. Melting
can also compute the thermodynamic of patterns with several adjacent GU base pairs.
This program computes the hybridisation enthalpy and entropy from the elementary
parameters of each Crick’s pair containing the GU base pairs.

∆hpattern−composed−of−GU−base−pairs = ∑ δ hCrick0 spair−containing−GU−base−pairs

Examples : One GU base pair
GUC GU UC
     
∆H = ∆H + ∆H
CGG CG GG

Examples : Two adjacent GU base pairs

GUGC GU UG UC
       
∆H = ∆H + ∆H + ∆H
CGUG CG GU GG

33
 (The same computation is performed for ∆S)
For further information, see the referenced articles.

model limits Article

tur99 RNA Mathiews et al (1999)
J.Mol.Biol. 288: 911-940
model limits Article
ser12 RNA Serra et al (2012)
Biochemistry 51: 3508-3522

Figure 12: Comparison of experimental and computed Tm for various sets of RNA
sequences containing GU base pairs. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

4.1.7 Single dangling end effect

The single dangling ends, that is the unmatched terminal nucleotides, can be taken into
account, but all the thermodynamic parameters are not available. In such a case, the
result is unpredictable, and all cases are possible.
For DNA and RNA duplexes, this program computes the hybridisation enthalpy
and entropy from the elementary parameters of the Crick’s pair containing the single
dangling end.

∆hsingle−dangling−end = δ hCrick0 s−pair−containing−the−dangling−end

Example : If the duplex is :

GCTAG-
CGATCA

34
 GCTAG-
 
∆H = ∆Hperfectly-matching-sequence + ∆Hsingle-dangling-end
CGATCCA
GCTAG- GCTAG G-
     
∆H = ∆H + ∆H
CGATCCA CGATCC CA

(The same computation is performed for ∆S)

For further information, see the referenced articles.

model limits Article

bom00 DNA Bommarito et al. (2000)
Nuc Acids Res 28: 1929-1934
sugdna02 DNA Ohmichi et al. (2002)
only terminal J. Am. Chem. Soc. 124: 10367-10372
poly A
self complementary
sequences
sugrna02 RNA Ohmichi et al. (2002)
only terminal poly A J. Am. Chem. Soc. 124: 10367-10372
self complementary
sequences
ser08 RNA O’tool et al. (2006)
only 3’ UA, Nucleic Acids research 34: 3338-3344
GU and UG terminal Miller et al. (2008)
base pairs Nucleic Acids research 36: 5652-5659
only 5’ UG and GU
terminal base pairs

35
Figure 13: Comparison of experimental and computed Tm for various sets of DNA
sequences containing single dangling ends. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

Figure 14: Comparison of experimental and computed Tm for various sets of RNA
sequences containing single dangling ends. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

36
4.1.8 Double dangling end effect
The double dangling ends, that is the two adjacent unmatched terminal nucleotides,
can be taken into account (mostly for RNA sequences). This program computes the
hybridisation enthalpy and entropy in two times : First, it computes the energy from
the single dangling end as if the duplex contained only a single danging end and then,
it adds a bonus for the second dangling end if it is necessary.

∆hdouble−dangling−end = δ hsingle−dangling−end
+ δ hbonus−second−dangling−end

Example :

UAC
 
∆H = ∆HUA + ∆Hbonus-pyrimidine-purine-pyrimidine
A

(The same computation is performed for ∆S)

For further information, see the referenced articles.

model limits Article

sugdna02 DNA Ohmichi et al. (2002)
only terminal J. Am. Chem. Soc. 124: 10367-10372
poly A
self complementary
sequences
sugrna02 RNA Ohmichi et al. (2002)
only terminal J. Am. Chem. Soc. 124: 10367-10372
poly A
self complementary
sequences
ser05 RNA O’toole et al. (2005)
depends on RNA 11: 512-516
the available
thermodynamic
parameters for
single dangling
ends
ser06 RNA O’toole et al. (2006)
Nucleic Acids research 34: 3338-3344

37
Figure 15: Comparison of experimental and computed Tm for various sets of RNA
sequences containing double dangling ends. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

4.1.9 Long dangling end effect (poly A)

The long dangling ends, that is all the adjacent unmatched terminal nucleotides, can
be taken into account (only for polyA dangling ends for the moment). It is possible to
compute the thermodynamic form one to four poly A dangling end. This program com-
putes the hybridisation enthalpy and entropy from the parameters of the long dangling
end with the adjacent terminal base pair.

∆hlong−dangling−end = δ hadjacent−terminal−base−pair+polyA

Example :
If the duplex is :
GCTAG---
CGATCAAA

38
 GCTAG-
 
∆H = ∆Hperfectly-matching-sequence + ∆Hlong-dangling-end
CGATCCAAA
GCTAG- GCTAG G-
     
∆H = ∆H + ∆H
CGATCCAAA CGATCC CAAA

(The same computation is performed for ∆S)

For further information, see the referenced articles.

model limits Article

sugdna02 DNA Ohmichi et al. (2002)
only terminal J. Am. Chem. Soc. 124: 10367-10372
poly A
self complementary
sequences
sugrna02 RNA Ohmichi et al. (2002)
only terminal J. Am. Chem. Soc. 124: 10367-10372
poly A
self complementary
sequences

Figure 16: Comparison of experimental and computed Tm for various sets of DNA
sequences containing long polyA dangling ends. [Na+ ] = 1 M, [nucleic acid] = 1 ·
10−4 M

39
Figure 17: Comparison of experimental and computed Tm for various sets of RNA
sequences containing long polyA dangling ends. [Na+ ] = 1 M, [nucleic acid] = 1 ·
10−4 M

40
4.1.10 Single bulge loop effect
The single bulge loops, that is the single unmatched internal nucleotides, can be taken
into account. , but all the thermodynamic parameters are not available. In such a case,
the result is unpredictable, and all cases are possible. There are several different models
to compute the thermodynamic of single bulge loop:

DNA and RNA duplexes :nearest neighbor model "NNN"

∆h(single-bulge-loop) = δ hunpaired−nucleotid+adjacent−base−pairs

Example : If the duplex is :

GCTTAGGC
CGA-TCCG

GCTTAGGC
 
∆H = ∆Hperfectly-matching-sequence-1 + ∆Hsingle-bulge-loop
CGA-TCCG
+ ∆Hperfectly-matching-sequence-2
GCTTAGGC GCT TTA AGGC
       
∆H = ∆H + ∆H + ∆H
CGA-TCCG CGA A-T TCCG

(The same computation is performed for ∆S)

However, some types of single bulge loop can’t be only modelled with a NNN
nearest neighbor model and the following models can give more reliable and accurate
results (mostly for RNA single bulge loops.)

DNA duplexes :Model from John Santalucia, Jr. and Donald Hicks, 2004

∆h(single-bulge-loop) = δ hintervening−NN + δ hclosing−AT−penalty

∆s(single-bulge-loop) = δ sbulge−loop−of−1 + δ sintervening−NN + δ sclosing−AT−penalty

Where :
δ hbulge−loop−of−1 accounts for the bulge loop of 1 nucleotide.
δ hintervening−NN accounts for the intervening base pair stack.
δ hclosing−AT−penalty accounts for each AT base pair adjacent to the single bulge loop.
Example :

GAC GC
   
∆H = ∆H
C-G CG
GAC GC
   
∆S = ∆Sbulge-loop-of-1 + ∆S
C-G CG

41
 (The same computation is performed for ∆S)

RNA duplexes :Model from Zhi Johm Lu, Douglas H. Turner and David H. Math-
ews, 2006

∆h(single-bulge-loop) = δ hinitiation−bulge−loop−of−1 + δ hintervening−NN

Where :
δ hinitiation−bulge−loop−of−1 accounts for the initiation of bulge loop of 1 nucleotide.
δ hintervening−NN accounts for the intervening base pair stack.
Example :

GAC GC
   
∆H = ∆Hinitiation-bulge-loop-of-1 + ∆H
C-G CG

(The same computation is performed for ∆S) For further information, see the refer-
enced articles.

model limits Article

tan04 DNA Tanaka et al (2004)
Biochemistry 43 : 7143-7150
san04 DNA Santalucia and Hicks (2004)
missing Annu. Rev. Biophys. Biomol. Struct 33 : 415-440
closing AT
penalty
ser07 RNA Blose et al (2007)
les reliable Biochemistry 46 : 15123-15135
results
some missing
parameters
tur06 RNA Lu et al (2006)
Nucleic Acids Research 34: 4912-4924

42
Figure 18: Comparison of experimental and computed Tm for various sets of DNA
sequences containing one single bulge loop. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

Figure 19: Comparison of experimental and computed Tm for various sets of RNA
sequences containing one single bulge loop. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

43
4.1.11 long bulge loop effect
The long bulge loops, that is all the adjacent unmatched internal nucleotides, can be
taken into account. , but all the thermodynamic parameters are not available. In such
a case, the result is unpredictable, and all cases are possible. The RNA and DNA
thermodynamic models are similar :

DNA duplexes :Model from John Santalucia, Jr. and Donald Hicks, 2004

∆h(long-bulge-loop) = δ hclosing−AT−penalty
∆s(single-bulge-loop) = δ sbulge−loop−of−n + δ sclosing−AT−penalty

Where :
δ hbulge−loop−of−n accounts for the bulge loop of n nucleotides.
δ hclosing−AT−penalty accounts for each AT base pair adjacent to the long bulge loop.
Example :

GACGC GACGC
   
∆H = 0∆S = ∆Sbulge-loop-of-3
C-G C-G

RNA duplexes : Model from Zhi Johm Lu, Douglas H. Turner and David H. Math-
ews, 2006

∆h(long-bulge-loop) = δ hinitiation−bulge−loop−of−n
+ δ hper−AU/GU−penalty

Where :
δ hinitiation−bulge−loop−of−n accounts for the initiation of the bulge loop of n nu-
cleotides.
δ hper−AU/GU−penalty accounts for each AU or GU base pair adjacent to the long
bulge loop.
Example :

AACGC
 
∆H = ∆Hinitiation-bulge-loop-of-3 + 1 × ∆Hper-AU-penalty
U-G

(The same computation is performed for ∆S)

For further information, see the referenced articles.

44
 model limits Article
san04 DNA Santalucia and Hicks (2004)
missing closing Annu. Rev. Biophys. Biomol. Struct 33 : 415-440
AT penalty
not tested
with experimental
results
tur06 RNA Lu et al (2006)
not tested Nucleic Acids Research 34: 4912-4924
with experimental
results

4.1.12 Inosine bases effect

The inosine bases (I) are taken into account, but all the thermodynamic parameters are
not available. In such a case, the result is unpredictable, and all cases are possible,
so the program quit with a warning. For the RNA duplexes, only the thermodynamic
parameters for IU base pairs are available for the moment. This program computes
the hybridisation enthalpy and entropy from the elementary parameters of each Crick’s
pair containing the inosine base.

∆hpattern−containing−inosine−bases = ∑ δ hCrick0 spair−containing−inosine−bases

Examples : One inosine base

AIC AI IC
     
∆H = ∆H + ∆H
TAG TA AG

Examples : Two adjacent base pairs containing inosine

GIAC GI IA AC
       
∆H = ∆H + ∆H + ∆H
CAIG CA AI IG

(The same computation is performed for ∆S)

For further information, see the referenced articles.

45
 model limits Article
san05 DNA Watkins and Santalucia (2005)
missing parameters Nucleic acids research 33 : 6258-6267
for tandem
base pairs
containing
inosine bases
zno07 RNA Wright et al. (2007)
only IU base Biochemistry 46 : 4625-4634
pairs

Figure 20: Comparison of experimental and computed Tm for various sets of DNA
sequences containing inosine. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

46
Figure 21: Comparison of experimental and computed Tm for various sets of RNA
sequences containing inosine. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

47
4.1.13 Azobenzenes effect
The trans azobenzenes (X_T) and cis azobenzenes (X_C) in DNA duplexes are taken
into account. Be aware : when the number of cis azobenzenes increases in the se-
quence, the predictions are less accurate and less reliable.

∆hpattern−containing−azobenzene = δ hCrick0 spair−containing−azobenzene+adjacent−base−pairs

Example : If the duplex is :

GCTX_CAGGC
CGATCCG

GCTX_CAGGC
 
∆H = ∆Hperfectly-matching-sequence-1 + ∆Hazobenzene
CGATCCG
+ ∆Hperfectly-matching-sequence-2
GCTX_CAGGC GCT TX_CA AGGC
       
∆H = ∆H + ∆H + ∆H
CGATCCG CGA AT TCCG

(The same computation is performed for ∆S)

For further information, see the referenced articles.

model limits Article

asa05 DNA Asanuma et al. (2005)
less reliable Nucleic acids Symposium Series 49 : 35-36
results when
the number
of cis azobenzene
increases

48
Figure 22: Comparison of experimental and computed Tm for various sets of DNA
sequences containing azobenzene. [Na+ ] = 1 M, [nucleic acid] = 2 · 10−6 M

4.1.14 2-Hydroxyadenine bases effect

The 2-hydroxyadenine bases (A*) in DNA duplexes are taken into account, but only
in this two different sequence contexts : 5’ GA*C 3’ and 5’ TA*A 3’. The program
computes the enthalpy and the entropy in two times : first it computes the enthalpy
and entropy of the two Crick’s pairs containing the hydroxyadenine as if the base pair
containing the hydroxyadenine was a simple AT base pair, and then it computes the
hydroxyadenine increments.

∆hpattern−containing−hydroxyadenine = ∑ δ hCrick0 spair−with−AT−base−pair

+ δ hhydroxyadenine−increment

Examples

GA*C GA AC
     
∆H = ∆H + ∆H
CCG CT CG
+ ∆Hincrement-for-GA*C/CCG

(The same computation is performed for ∆S)

49
 For further information, see the referenced articles.

model limits Article

sug01 DNA Kawakami et al.(2001)
only in 5’ Nucleic acids research 29 : 3289-3296
GA*C 3’
and 5’ TA*A
contexts

Figure 23: Comparison of experimental and computed Tm for various sets of DNA
sequences containing hydroxyadenine. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

50
4.1.15 Single Locked nucleic acid effect
The locked nucleic acids (AL, GL, CL, TL) in DNA duplexes are taken into account.

DNA duplexes : Model from McTigue et al., 2004 The program computes the
enthalpy and the entropy in two times : first it computes the enthalpy and entropy of
the two Crick’s pairs containing the locked nucleic acid as if the locked nucleic acid
was a simple nucleic acid, and then it computes the locked nucleic acid increments for
each Crick’s base pair containing the locked nucleic acid.

∆hpattern−containing−Locked−Nucleic−Acid = ∑ δ hCrick0 spair−without−Locked−Nucleic−Acid

+ ∑ δ hincrement−Crick0 spair−with−Locked−Nucleic−Acid

Examples

GALC GA AC
     
∆H = ∆H + ∆H
CTG CT CG
+ ∆Hincrement-for-GAL/CT + ∆Hincrement-for-ALC/TG

(The same computation is performed for ∆S)

DNA duplexes : Model from Owczarzy et al., 2011 The program computes the
enthalpy and the entropy following the same nearest-neighbor formula as for perfectly
matching Crick’s pairs.
Examples

GALC GAL ALC

     
∆H = ∆H + ∆H
CTG CT CG

For further information, see the referenced articles.

model limits Article

mct04 DNA McTigue et al.(2004)
Biochemistry 43 : 5388-5405
owc11 DNA Owczarzy et al.(2011)
Biochemistry 50 : 9352-9367

51
Figure 24: Comparison of experimental and computed Tm for various sets of DNA
sequences containing single Locked Nucleic Acid. [Na+ ] = 1 M, [nucleic acid] = 5 ·
10−6 M

4.1.16 Consecutive Locked nucleic acids effect

The consecutive locked nucleic acids (AL, GL, CL, TL) in DNA duplexes are taken
into account. The program computes the enthalpy and the entropy following the same
nearest-neighbor formula as for perfectly matching Crick’s pairs.
Examples

GALCLC GAL ALCL CLC

       
∆H = ∆H + ∆H + ∆H
CTGG CT TG GG

For further information, see the referenced articles.

model limits Article

owc11 DNA Owczarzy et al.(2011)
Biochemistry 50 : 9352-9367

52
Figure 25: Comparison of experimental and computed Tm for various sets of DNA se-
quences containing consecutive Locked Nucleic Acids. [Na+ ] = 1 M, [nucleic acid] =
2 · 10−6 M

4.1.17 Consecutive Locked nucleic acids with a single mismatch effect

The consecutive locked nucleic acids with a single mismatch (AL, GL, CL, TL) in
DNA duplexes are taken into account. The program computes the enthalpy and the
entropy following the same nearest-neighbor formula as for perfectly matching Crick’s
pairs.
Examples

GALCLGLC GAL ALCL CLGL GLC

         
∆H = ∆H + ∆H + ∆H ∆H
CTTCG CT TT TC CG

For further information, see the referenced articles.

model limits Article

owc11 DNA Owczarzy et al.(2011)
Biochemistry 50 : 9352-9367

53
Figure 26: Comparison of experimental and computed Tm for various sets of DNA se-
quences containing consecutive Locked Nucleic Acids and a single mismatch. [Na+ ] =
1 M, [nucleic acid] = 2 · 10−6 M

4.2 The melting temperature

Then the melting temperature is computed by the following formula:
Tm = ∆S+R ∆H ln(CT /F) − 273.15

Tm in K (for [Na+ ] = 1 M)
In case of self complementary sequences, if the sequence (5’ 3’) is a sequence
of type G(CNG)xC and x > 4, the sequence mainly turns into hairpin loops and this
program will compute the melting temperature with this formula :
Tm = ∆H ∆S − 273.15

Tm in K
Moreover, no ion correction will be applied to this formula.

4.3 Correction for the concentration of nucleic acid

F is 1 in the case of self-complementarity oligonucleotides. If the ODNs are not self-
complementary, F is 4 if both strands are present in equivalent amount and F is 1 if
one strand is in excess (for instance in PCR experiments). As a matter of facts, when
the oligonucleotides are not self-complementary, the formula in the denominator is
Cmax − Cmin /2 where Cmax is the concentration of the strand in excess and Cmin
the concentration of the other strand. If the excess is sufficient, the total concentration
is equivalent to the concentration of the strand in excess. Therefore, if one strand
is in excess, the actual formula is effectively Cmax − Cmin /2 but if Cmax  Cmin ,

54
Cmax −Cmin /2 is equivalent to the total concentration CT . If Cmax is close to Cmin ,
Cmax −Cmin /2 is equivalent to CT /4, which is the default correction.
F is 4 by default but note that MELTING can detect self complementary sequences
for perfectly matching sequences even though there is(are) dangling end(s). In this
case, the program will automatically change F to 1. In addition to that, the compu-
tation takes an entropic term to correct for self-complementarity. In case of other self
complementary sequences which doesn’t match perfetcly, the option -self must be used
to inform the program of the self complementarity.

Figure 27: Comparison of experimental and computed Tm for various sets of DNA self
complementary sequences. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

55
Figure 28: Comparison of experimental and computed Tm for various sets of RNA self
complementary sequences. [Na+ ] = 1 M, [nucleic acid] = 1 · 10−4 M

56
4.4 Correction for the concentration of cations
After the program computed the melting temperature for [Na+ ]=1, an ion correction
wille be applied either directly on the computed melting temperature or on the com-
puted entropy. In the last case, the melting temperature is computed using the first
formula of the Melting temperature section. We must enter at least one of the follow-
ing ion concentrations : [Na+ ], [K+ ], [Tris+ ] or [Mg2+ ] and several ion corrections are
proposed (see the reference table to have more information):

4.4.1 Sodium corrections

• ahs01
∆S = ∆S[Na+ ]=1 M + 0.847 × (N − 1) × log[Na+ ]
Where N is the length of the duplex.
• kam71
T m = T m[Na+ ]=1 M + (7.95 − 3.06 × χGC ) × ln[Na+ ]
Where χGC is the frequence of GC base pairs in the duplex.
• marschdot

T m = T m[Na+ ]=1 M + (8.75 − 2.83 × χGC ) × ln[Na+ ]

Where χGC is the frequence of GC base pairs in the duplex.

• owc1904

T m = T m[Na+ ]=1 M + (−3.22 × χGC − 6.39) × ln[Na+ ]

Where χGC is the frequence of GC base pairs in the duplex.

• owc2004
1 1 1
= + (3.85 × χGC − 6.18) × × ln[Na+ ]
T m T m[Na+ ]=1 M 100000

Where χGC is the frequence of GC base pairs in the duplex.

• owc2104

T m = T m[Na+ ]=1 M + (−4.62 × χGC + 4.52) × ln[Na+ ]

−0.985 × ln[(Na+ )2 ]
Where χGC is the frequence of GC base pairs in the duplex.

57
• owc2204

1 1
= + (4.29 × χGC − 3.95)×
Tm T m[Na+ ]=1 M

1 1
× ln[Na+ ] + 9.40 × × ln[Na+ ]
100000 1000000
Where χGC is the frequence of GC base pairs in the duplex.
• san96
12.5 log[Na+ ]

• san04
∆S = ∆S[Na+ ]=1 M + 0.368 ∗ (N − 1) × ln[Na+ ]
Where N is the length of the duplex.
• schlif
T m = T m[Na+ ]=1 M + 16.6 × log[Na+ ]

• tanna06
∆S = ∆S[Na+ ]=1 M − 3.22 × (N − 1) × g1
Where N is the length of the duplex.

b1
g1 = a1 +
N

a1 = −0.07 × ln[Na+ ] + 0.012 × (ln[Mg2+ ])2

b1 = 0.013 × (ln[Mg2+ ])2
item tanna07
∆S = ∆S[Na+ ]=1 M − 3.22 × (N − 1) × g1
Where N is the length of the duplex.

b1
g1 = a1 +
N

a1 = −0.075 × ln[Na+ ] + 0.012 × (ln[Mg2+ ])2

b1 = 0.018 × (ln[Mg2+ ])2

• wet91
[Na+ ]
T m = T m[Na+ ]=1 M + 16.6 log + 3.85
1 + 0.7[Na+ ]

58
correction contexts Article
ahs01 DNA von Ahsen et al ,2001
Na>0 Clinical Chemistry, 47, 1956-1961.
kam71 DNA Frank-Kamenetskii et al. 1971
Na>=0.069 Biopolymers 10, 2623-2624.
Na<=1.02
marschdot DNA Marmur, J., and Doty, P. (1962)
Na>=0.069 J. Mol. Biol. 5, 109-118.
Na<=1.02 Blake and Delcourt. (1998) Nucleic Acids Res. 26, 3323-3332
and corrigendum.
owc1904 DNA Owczarzy et al.,2004
Na>0 Biochemistry,43, 3537-3554.
owc2004 DNA Owczarzy et al., 2004
Na>0 Biochemistry, 43, 3537-3554.
owc2104 DNA Owczarzy et al., 2004
Na>0 Biochemistry, 43, 3537-3554.
owc2204 DNA Owczarzy et al., 2004
Na>0 Biochemistry, 43, 3537-3554.
owc2204 DNA Owczarzy et al., 2004
Na>0 Biochemistry,43, 3537-3554.
san96 DNA SantaLucia et al.(1996)
Na>=0.1 Biochemistry 35 : 3555-3562
san04 DNA Santalucia and Hicks (2004)
Na>=0.05 Annu. Rev. Biophys. Biomol. Struct 33 : 415-440
Na<=1.1 John Santalucia, Jr., 1998
Proc. Natl. Acad. Sci. USA, 95, 1460-1465
oligonucleotides
inferior to 16 bases
schlif DNA Schildkraut, C., and Lifson, S. (1965)
Na>=0.07 Biopolymers 3, 195-208.
Na<=0.12
tanna06 DNA Tan et al. 2006,
Na>=0.001 Biophysical Journal, 90, 1175-1190.
Na<=1
tanna07 RNA Tan et al, 2007
Na>=0.003 Biophysical Journal, 92, 3615-3632.
Na<=1
wet91 RNA, DNA Wetmur 1991
and RNA/DNA Critical reviews in biochemistry and molecular
Na>0 biology, 26, 227-259

59
4.4.2 Magnesium corrections
• owcmg08
1 1 1
= +a−b(ln[Mg2+ ])+ χGC (c+d ln[Mg2+ ])+
T m[Mg2+ ] T m[Na+ ]=1 M 2(Nbp − 1)

(−e + f ln[Mg2+ ] + g(ln[Mg2+ ])2 ])

Where χGC is the frequence of GC base pairs in the duplex. Nbp is the number
of base pairs and a, b, c, d, e, f, g fixed to :
a = 3.92 x 10−5
b = 9.11 x 10−6
c = 6.26 x 10−5
d = 1.42 c 10−5
e = 4.82 x 10−4
f = 5.25 x 10−4
g = 8.31 x 10−5 .

• tanmg06
∆S = ∆S[Na+ ]=1 M − 3.22 × (N − 1) × g2
Where N is the length of the duplex.

b2
g2 = a2 +
(N)2

a2 = 0.02 × ln[Mg2+ ] + 0.0068 × (ln[Mg2+ ])2

b2 = 1.18 × ln[Mg2+ ] + 0.344 × (ln[Mg2+ ])2
item tanmg07
∆S = ∆S[Na+ ]=1 M − 3.22 × (N − 1) × g2
Where N is the length of the duplex.

b2
g2 = a2 +
(N)2

−0.6
a2 = + 0.025 × ln[Mg2+ ] + 0.0068 × (ln[Mg2+ ])2
N
b2 = ln[Mg2+ + 0.38 × (ln[Mg2+ ])2

60
 correction limits Article
oxcmg08 DNA Owczarzy et al.,2008
Mg>=0.0005 Biochemistry, 47, 5336-5353.
Mg<=0.6
tanmg06 DNA Tan et al. 2006
Mg>=0.0001 Biophysical Journal, 90, 1175-1190.
Mg<=1
oligomer
length
superior to
6 base pairs
tanmg07 RNA Tan et al, 2007
Mg>=0.1 Biophysical Journal, 92, 3615-3632.
Mg<=0.3

4.4.3 Mixed Na Mg corrections

• owcmix08
1 1 1
= +a−b(ln[Mg2+ ])+ χGC (c+d ln[Mg2+ ])+
T m[Mg2+ ] T m[Na+ ]=1 M 2(Nbp − 1)

(−e + f ln[Mg2+ ] + g(ln[Mg2+ ])2 ])

Where χGC is the frequence of GC base pairs in the duplex. Nbp is the number
of base pairs. b, c, e, f are fixed as in the magnesium correction owcmg08.
a = 3.92 · 10−5 (0.843 − 0.352[Mon+ ]0.5 ln[Mon+ ])
d = 1.42 · 10−5 (1.279 − 4.03 · 10−3 ln[Mon+ ] − 8.03 · 10−3 ln[Mon+ ]2 )
g = 8.31 · 10−5 (0.486 − 0.258 ln[Mon+ ] + 5.25 · 10−3 ln[Mon+ ]3
• tanmix07
∆S = ∆S[Na+ ]=1 M − 3.22 × ((N − 1) × (x1 × g1 + x2 × g2) + g12))

Where N is the length of the duplex.

1
ln[( x1 − 1) × Na+ ]
g12 = −0.6 × x1 × x2 × ln[Na+ ] ×
N
See what is g1 and g2 in the sodium corrections tanna06 and tanna07 (g1) and
magnesium corrections tanmg06 and tanmg07 (g2). Formula representing the
fractional contribution of Na+ ions.
[Na+ ]
x1 =
(Na+ + ( 8.1−32.4
N ) × (5.2 − ln[Na+ ]) × Mg2+ )
Formula representing the fractional contribution of Mg2+ ions.
x2 = 1 − x1

61
 correction limits Article
oxcmix08 DNA Owczarzy et al.,2008
Mg>=0.0005 Biochemistry, 47, 5336-5353.
Mg<=0.6
Na+K+Tris/2>0
tanmix07 DNA Tan et al, 2007
and RNA Biophysical Journal, 92, 3615-3632.
Mg>=0.1
Mg<=0.3
Na+K+Tris/2>=0.1
Na+K+Tris/2<=0.3

If the user doesn’t enter any ion correction, the algorithm from Owczarzy et al.
(2008) will be used by default :

[Mon+ ] = [Na+ ] + [k+ ] + [Tris+ ]

Where [Tris+ ] is equal to half of total tris buffer concentration. (in the option -t, it is
the Tris buffer concentration which is entered).

• if [Mon+ ] = 0, a default sodium correction will be used.

• if [Mg2+ ]0 .5 / [Mon+ ] < 0.22, a default sodium correction is used. Monovalent
ion influence is dominant, divalent cations can be disregarded.

• if [Mg2+ ]0 .5 / [Mon+ ] >= 0.22 and [Mg2+ ]0 .5 / [Mon+ ] < 6, a default mixed Na
Mg correction is used. We can have a competitive DNA or RNA binding between
monovalent and divalent cations.
• if [Mg2+ ]0 .5 / [Mon+ ] >= 6, a default magnesium correction is used. Divalent
cation influence is dominant, monovalent cations can be disregarded.

Moreover, if the user wants to use a sodium correction but also enters a potas-
sium, Tris buffer and/or a magnesium concentration, a sodium equivalent concentra-
tion which takes into account the other ion concentrations is computed before applying
the sodium correction. Several sodium equivalence ready to use are proposed by this
program :
• ahs01
[Tris+ ]
q
+ + +
[NaEq ] = [Na ] + [K ] + + 3.79 [Mg2+ ] − [dNTP]
2

• mit96
[Tris+ ]
q
[NaEq+ ] = [Na+ ] + [K+ ] + + 4 [Mg2+ ] − [dNTP]
2

62
 • pey00

[Tris+ ]
q
[NaEq+ ] = [Na+ ] + [K+ ] + + 3.3 [Mg2+ ] − [dNTP]
2

For further information, see the referenced articles :

correction limits Article

ahs01 DNA von Ahsen et al. 2001
Clinical Chemistry, 47, 1956-1961.
mit96 DNA Mitsuhashi. et al, 1996
J. Clin. Lab. Anal, 10, 277-284.
pey00 DNA Peyret, 2000
Ph.D Thesis, Section .5.4.2, 128, Wayne State
University, Detroit, MI

4.5 Correction for the concentration of denaturing agents

MELTING is currently accurate when the hybridisation is performed at pH 7 ± 1, but
some temperature corrections for the formamide and DMSO concentrations exists and
can be applied. However, these corrections are rough approximations and results accu-
racy may be lost.

4.5.1 DMSO corrections, DMSO in %

• ahs01
T m = T m(DMSO = 0) − 0.75 × DMSO

• cul76
T m = T m(DMSO = 0) − 0.5 × DMSO

• esc80
T m = T m(DMSO = 0) − 0.675 × DMSO

• mus81
T m = T m(DMSO = 0) − 0.6 × DMSO

For further information, see the referenced articles :

63
 correction limits Article
ahs01 DNA von Ahsen et al. 2001
not tested Clinical Chemistry, 47, 1956-1961.
with experimental
values
cul76 DNA Cullen et al., 1976
not tested 3, 49-62.
with experimental
values
esc80 DNA Escara et al., 1980
not tested 19, 1315-1327.
with experimental
values
mus80 DNA Musielski et al., 1981
not tested Z allg Microbiol 1981; 21, 447-456.
with experimental
values

4.5.2 formamide corrections

• bla96

T m = T m( f ormamide = 0) + (0.453 × χGC − 2.88) × f ormamide

Where χGC is the frequence of GC base pairs in the sequence. formamide is in

mol/L
• lincorr
T m = T m( f ormamide = 0) − 0.65 × f ormamide
Where formamide is in %.

For further information, see the referenced articles :

64
 correction limits Article
bla96 DNA Blake and Delcourt, 1996
not tested Vol. 24, No. 11 2095-2103
with experimental
values
formamide in mol/L
lincorr DNA McConaughy et al., 1969
not tested Biochemistry 8, 3289-3295.
with experimental Record, M.T., Jr, 1967
in % Biopolymers, 5, 975-992.
values Casey et al, 1977
Formamide in Nucleic acids research, 4, 1539-1532.
% Hutton, 1977
Nucleic acids research, 4, 3537-3555.

4.6 Long sequences

It is important to realise that the nearest-neighbor approach has been established on
small oligonucleotides. Therefore the use of MELTING in the non-approximative mode
is really accurate only for relatively short sequences (Although if the sequences are
two short, let’s say < 6 bp, the influence of extremities becomes too important and
the reliability decreases a lot). For long sequences an approximative mode has been
designed. This mode is launched if the sequence length is higher than the value given
by the option -T (the default threshold is 60 bp).
The melting temperature can be computed by one of the following formulas:
• ahs01
550
T m = 80.4 + 0.345 × %GC + log[Na+ ] × (17.0 − 0.135 × %GC) −
size

• che93
650
T m = 69.3 + 0.41 × %GC −
size
• che93corr
535
T m = 69.3 + 0.41 × %GC −
size
• marschdot
675
T m = 81.5 + 16.6 × log[Na+ ] + 0.41 × %GC −
size

• owe69

T m = 87.16 + 0.345 × %GC + log[Na+ ] × (20.17 − 0.066 × %GC)

65
 • san98
528
T m = 77.1 + 11.7 × log[Na+ ] + 0.41 × %GC −
size
• wetdna91
[Na+ ] 500
T m = 81.5 + 16.6 log + 0.41%GC − − %Mismatching
1 + 0.7[Na+ ] size

• wetrna91
[Na+ ] 500
T m = 78 + 16.6 log + + 0.7%GC − − %Mismatching
1 + 0.7[Na ] size

• wetdnarna91
[Na+ ] 500
T m = 67 + 16.6 log + + 0.8%GC − − %Mismatching
1 + 0.7[Na ] size

For further information, see the referenced articles :

Figure 29: Comparison of experimental and computed Tm for various sets of DNA
approximative formulas. [Na+ ] = 1 M

66
Figure 30: Comparison of experimental and computed Tm for various sets of RNA
approximative formulas. [Na+ ] = 1 M

Figure 31: Comparison of experimental and computed Tm for various sets of

DNARNA approximative formulas. [Na+ ] = 1 M, [nucleic acid] = 4 · 10−4 M

67
 formula limits Article
ahs01 DNA von Ahsen et al. 2001
no mismatch Clinical Chemistry, 47, 1956-1961.
che93 DNA Marmur et al., 1962
no mismatch Journal of molecular biology, 5, 109-118.
Na=0 Chester N et al. 1993
Mg=0.0015 Analytical Biochemistry, 209, 284-290.
Tris=0.01
K=0.05
che93corr DNA Marmur et al., 1962
no mismatch Journal of molecular biology, 5, 109-118.
Na=0 Chester N et al. 1993
Mg=0.0015 Analytical Biochemistry, 209, 284-290.
Tris=0.01 Nicolas Von Ahsen et al. 2001
K=0.05 Clinical Chemistry, 47, 1956-1961.
marschdot DNA Wetmur,1991
no mismatch Critical reviews in biochemistry
and molecular biology, 26, 227-259
Marmur et al., 1962
Journal of molecular biology, 5, 109-118.
Chester et al., 1993
Analytical Biochemistry, 209, 284-290.
Schildkraut et al., 1965
Biopolymers, 3, 95-110.
Wahl et al., 1987
Methods Enzymol;152:399 - 407.
Britten et al.,1974
Methods Enzymol ;29:363-418.
Hall et al., 1980
J Mol Evol ;16:95-110.
owe69 DNA Owen et al., 1969
no mismatch Biopolymers, 7:503-16.
Frank-Kamenetskii,1971
Biopolymers;10:2623- 4.
Blake, 1996
Encyclopedia of molecular biology and
molecular medicine, Vol. 2., :1-19.
Blake et al.,1998
Nucleic Acids Res;26:3323-32.
san98 DNA Santalucia, 1998
no mismatch Proc Nacl Acad Sci USA
95, 1460-1465.
von Ahsen et al. 2001,
Clinical Chemistry, 47, 1956-1961.
wetdna91 DNA Wetmur,1991,
Critical reviews in biochemistry
and molecular biology, 26, 227-259
wetrna91 RNA 68 Wetmur,1991,
Critical reviews in biochemistry
and molecular biology, 26, 227-259
wetdnarna91 DNA/RNA Wetmur,1991,
Critical reviews in biochemistry
and molecular biology, 26, 227-259
References
[1] H.T. Allawi and J. SantaLucia. Thermodynamics and NMR of internal G-T mis-
matches in DNA. Biochemistry, 36:10581–10594, 1997.
[2] H.T. Allawi and J. SantaLucia. Nearest neighbor thermodynamics of internal
A.C mismatches in DNA: sequence dependence and pH effects. Biochemistry,
37:9435–9444, 1998.
[3] H.T. Allawi and J. SantaLucia. Nearest neighbor thermodynamics parameters for
internal G.A mismatches in DNA. Biochemistry, 37:2170–2179, 1998.
[4] H.T. Allawi and J. SantaLucia. Thermodynamics of internal C.T mismatches in
DNA. Biochemistry, 26:2694–2701, 1998.
[5] Hiroyuki Asanuma, Daijiro Matsunaga, and Makoto Komiyama. Clear-cut photo-
regulation of the formation and dissociation of the DNA duplex by modified
oligonucleotide involving multiple azobenzenes. Nucleic acids Symposium Se-
ries, 49:35–36, 2005.
[6] Jaya Badhwar, Saradasri Karri, Cody K. Cass, Erica L. Wunderlich, and Brent M.
Znosco. Thermodynamic characterization of RNA duplexes containing naturally
occuring 1x2 nucleotide internal loops. Biochemistry, 46:14715–14724, 2007.
[7] R. D. Blake and S. G. Delcourt. Thermal stability of DNA. Nucleic Acids Res,
26:3323–3332 and corrigendum, 1998.
[8] R. D. Blake and Scott G. Delcourt. Thermodynamic effects of formamide on
DNA stability. Nucleic Acids Research, 24, 11:2095–2103, 1996.
[9] RD. Blake. Denaturation of DNA. In: Meyers RA, ed. Encyclopediaof molecular
biology and molecular medicine, 2:1–19, 1996.
[10] Joshua Blose, M. Michelle L. Manni, Kelly A. Klapec, Yukiko Stranger-Jones,
Allison C. Zyra, Vasiliy Sim, Chad A. Griffith, Jason D. Long, and Martin J.
Serra. Non-nearest-neighbor dependence of stability for RNA bulge loops based
on the complete set of Group I single nucleotide bulge loops. Biochemistry,
46:15123–15135, 2007.
[11] S. Bommarito, N. Peyret, and J. SantaLucia. Thermodynamic parameters for
DNA sequences with dangling ends. Nucleic Acids Res, 28:1929–1934, 2000.
[12] K.J. Breslauer, R. Frank, H. Blöcker, and Marky L.A. Predicting DNA duplex
stability from the base sequence. Proc Natl Acad Sci USA, 83:3746–3750, 1986.
[13] RJ Britten, DE Graham, and BR Neufeld. Analysis of repeating DNA sequences
by reassociation. Methods Enzymol, 29:363–418, 1974.
[14] N. Casey, J.and Davidson. Rates of formation and thermal stabilities of
RNA:DNA and DNA:DNA duplexes at high concentrations of formamide. Nu-
cleic acids research, 4:1539–1532, 1977.

69
[15] Jonathan L. Chen, Abigael L. Dishler, Scott D. Kennedy, Ilyas Yildirim, Biao Liu,
Douglas H. Turner, and Martin J. Serra. Testing the nearest neighbor model for
canonical RNA base pairs: Revision of GU parameters. Biochemistry, 51:3508–
3522, 2012.
[16] N Chester and DR Marshak. dimethyl sulfoxide-mediated primer tm reduction
: a method for analyzing the role of renaturation temperature in the polymerase
chain reaction. Analytical Biochemistry, 209:284–290, 1993.
[17] Br Cullen and Md Bick. Thermal denaturation of DNA from bromodeoxyuridine
substitued cells. Nucleic acids research, 3:49–62, 1976.
[18] Amber R. Davis and Brent M. Znosko. Thermodynamic characterization of single
mismatches found in naturally occurring RNA. Biochemistry, 46:13425–13436,
2007.
[19] Amber R. Davis and Brent M. Znosko. Thermodynamic characterization of nat-
urally occurring RNA single mismatches with G-U nearest neighbors. Biochem-
istry, 47:10178–10187, 2008.
[20] JF Escara and Jr Hutton. Thermal stability and renaturation of DNA in dimethyl
sulfoxidesolutions: acceleration of the renaturation rate. Biopolymers, 19:1315–
1327, 1980.
[21] M. D. Frank-Kamenetskii. Simplification of the empirical relationship between
melting temperature of DNA, its GC content and concentration of sodium ions in
solution. Biopolymers, 10:2623–2624, 1971.
[22] S.M. Freier, R. Kierzek, J.A. Jaeger, N. Sugimoto, M.H. Caruthers, T. Neilson,
and D.H. Turner. Biochemistry, 83:9373–9377, 1986.
[23] TJ Hall, JW Grula, EH Davidson, and RJ Britten. Evolution of sea urchin non-
repetitive DNA. J Mol Evol, 16:95–110, 1980.
[24] JR Hutton. Renaturation kinetics and thermal stability of DNA in aqueous solu-
tions of formamide and urea. Nucleic acids research, 4:3537–3555, 1977.
[25] Junji Kawakami, Hiroyuki Kamiya, Kyohko Yasuda, Hiroyoshi Fujiki, Hiroshi
Kasai, and Naoki Sugimoto. Thermodynamic stability of base pairs between
2-hydroxyadenine and incoming nucleotides as a determinant of nucleotide in-
corporation specificity during replication. Nucleic acids research, 29:3289–3296,
2001.
[26] Elzbieta Kierzek, David H. Mathews, Anna Ciesielska, Douglas H. Turner, and
Ryszard Kierzek. Nearest neighbor parameters for Watson Crick complementary
heteroduplexesformed between 2-O-methyl RNA and RNA oligonucleotides. Nu-
cleic acids research, 34:3609–3614, 2006.
[27] Zhi John Lu, Douglas H. Turner, and David H. Mathews. A set of nearest neigh-
bor parameters for predicting the enthalpy change of RNA secondary structure
formation. Nucleic Acids Research, 34:4912–4924, 2006.

70
[28] Broda Magdalena, Elbieta Kierzek, Zofia Gdaniec, Tadeusz Kulinski, and
Ryszard Kierzek. Thermodynamic stability of RNA structures formed by CNG
trinucleotide repeats. implication for prediction of RNA structure. Biochemistry,
44:10873–10882, 2005.
[29] J. Marmur and P. Doty. Determination of the base composition of deoxyribonu-
cleic acid from its thermal denaturation temperature. J. Mol. Biol., 5:109–118,
1962.
[30] David H. Mathews, Jeffrey Sabina, Michael Zucker, and Douglas H Turner. Ex-
panded sequence dependence of thermodynamic parameters improves predic-
tionof RNA secondary structure. J. Mol. Biol, 288:911–940, 1999.
[31] B.L. McConaughy, C.D. Laird, and B.I. McCarthy. Nucleic acid reassociation in
formamide. Biochemistry, 8:3289–3295, 1969.
[32] Patricia M. McTigue, Raymond J. Peterson, and Jason D. Kahn. Sequence-
dependent thermodynamic parameters for locked nucleic acid (LNA)DNA du-
plexformation. Biochemistry, 43:5388–5405, 2004.
[33] Stacy Miller, Laura E. Jones, Karen Giovannitti, Dan Piper, and Martin J. Serra.
Thermodynamic analysis of 5 and 3 single- and 3 double-nucleotide overhangs
neighboring wobble terminal base pairs. Nucleic Acids research, 36:5652–5659,
2008.
[34] M. Mitsuhashi. Technical report: Part 1. basic requirements for designing optimal
oligonucleotide probe sequences. J. Clin. Lab. Anal, 10:277–284, 1996.
[35] H. Musielski, W Mann, R Laue, and S Michel. Influence of dimethylsulfoxide on
transcription by bacteriophage T3-induced RNA polymerase. Z allg Microbiol,
21:447–456, 1981.
[36] Tatsuo Ohmichi, Shu-ichi Nakano, Daisuke Miyoshi, and Naoki Sugimoto. Long
RNA danglingend has large energetic contribution to duplex stability. J. Am.
Chem. Soc., 124:10367–10372, 2002.
[37] Amanda S. O’toole, Stacy Miller, Nathan Haines, M. Coleen Zink, and Martin J
Serra. Comprehensive thermodynamic analysis of 3’ double-nucleotide over-
hangs neighboring Watson-Crick terminal base pairs. Nucleic Acids research,
34:3338–3344, 2006.
[38] Amanda S. O’toole, Stacy Miller, and Martin J Serra. Stability of 3’ double
nucleotide overhangs that model the 3’ ends of siRNA. RNA, 11:512–516, 2005.
[39] R. Owczarzy, B.G. Moreira, Y. You, M.B. Behlke, and J.A. Walder. Predicting
stability of DNA duplexes in solutions containing magnesium and monovalent
cations. Biochemistry, 47:5336–5353, 2008.
[40] Richard Owczarzy, Yong You, Christopher L. Groth, and Andrey V. Tataurov.
Stability and mismatch discrimination of locked nucleic acid DNA duplexes. Bio-
chemistry, 50:9352–9367, 2011.

71
[41] Richard Owczarzy, Yong You, Bernardo G. Moreira, Jeffrey A. Manthey, Lingyan
Huang, Mark A. Behlke, and Joseph A. Walder. Effects of sodium ions on dna
duplex oligomers: Improved predictions of melting temperatures. Biochemistry,
43:3537–3554, 2004.
[42] RJ Owen, LR Hill, and SP. Lapage. Determination of DNA base compositions
from melting profiles in dilute buffers. Biopolymers, 7:503–516, 1969.
[43] N. Peyret. Prediction of nucleic acid hybridization : parameters and algorithms.
Ph.D Thesis 4.2, 128, Wayne State University, Detroit, MI., 5, 2000.
[44] N. Peyret, P.A. Seneviratne, H.T. Allawi, and J. SantaLucia. Nearest neighbor
thermodynamics and NMR of DNA sequences with internal A.A, C.C, G.G and
T.T mismatches. dependence and pH effects. Biochemistry, 38:3468–3477, 1999.
[45] M.T. Record, Jr. Electrostatic effects on polynucleotide transitions. I. Behavior at
neutral pH. Biopolymers, 5:975–992, 1967.
[46] J. SantaLucia. A unified view of polymer, dumbbell, and oligonucleotide DNA
nearest-neighbor thermodynamics. Proc Natl Acad Sci USA, 95:1460–1465,
1998.
[47] J. SantaLucia and Donald Hicks. The thermodynamics of DNA structural motifs.
Annu. Rev. Biophys. Struct, 33:415–440, 2004.

[48] J. SantaLucia, Jr, H.T. Allawi, and P.A. Seneviratne. Improved nearest-neighbor
parameters for predicting DNA duplex stability. Biochemistry, 35:3555–3562,
1996.
[49] C. Schildkraut and S. Lifson. Dependence of the melting temperature of DNA on
salt concentration. Biopolymers, 3:195–208, 1965.

[50] N. Sugimoto, M. Katoh, S. Nakano, T. Ohmichi, and M. Sasaki. RNA/DNA hy-

brid duplexes with identical nearest-neighbor base-pairs have identical stability.
FEBS Letters, 354:74–78, 1994.
[51] N. Sugimoto, S. Nakano, M. Katoh, A. Matsumura, H. Nakamuta, T. Ohmichi,
M. Yoneyama, and M. Sasaki. Thermodynamic parameters to predict stability of
RNA/DNA hybrid duplexes. Biochemistry, 34:11211–11216, 1995.
[52] N. Sugimoto, S. Nakano, M. Yoneyama, and K. Honda. Improved thermodynamic
parameters and helix initiation factor to predict stability of DNA duplexes. Nuc
Acids Res, 24:4501–4505, 1996.

[53] Zhi-Jie Tan and Shi-Jie Chen. Nucleic acid helix stability: effects of salt concen-
tration, cation valence and size, and chain length. Biophysical Journal, 90:1175–
1190, 2006.
[54] Zhi-Jie Tan and Shi-Jie Chen. Rna helix stability in mixed Na+/Mg2+ solutions.
Biophysical Journal, 92:3615–3632, 2007.

72
[55] Fumiaki Tanaka, Atsushi Kameda, Masahito Yamamoto, and Azuma Ohuchi.
Thermodynamic parameters based on a nearest neighbor model for DNA se-
quences with a single bulge loop. Biochemistry, 43:7143–7150, 2004.
[56] Nicolas von Ahsen, Carl T Wittwer, and Ekkehard Schutz. Oligonucleotide
melting temperatures under PCR conditions : deoxynucleotide triphosphate and
dimethyl sulfoxide concentrations with comparison to alternative empirical for-
mulas. Clinical Chemistry, 47:1956–1961, 2001.
[57] GM Wahl, SL Berger, and AR. Kimmel. Molecular hybridization of immobi-
lized nucleic acids: theoretical concepts and practical considerations. Methods
Enzymol, 152:399–407, 1987.

[58] N.E. Watkins and J. Santalucia, Jr. Nearest-neighbor thermodynamics of deoxyi-

nosine pairs in DNA duplexes. Nuc Acids Res, 33:6258–6267, 2005.
[59] Norman E. Watkins, Jr, William J. Kennely, Mike J. Tsay, Astrid Tuin, Lara
Swenson, Hyung-Ran Lee, Svetlana Morosyuk, Donald A. Hicks, and John
Santalucia, Jr. Thermodynamic contributions of single internal rA.dA, rC.dC,
rG.dG and rU.dT mismatches in RNA/DNA duplexes. Nucleic Acids Research,
39:1894–1902, 2011.
[60] J.G. Wetmur. DNA probes: applications of the principles of nucleic acid hy-
bridization. Crit Rev Biochem Mol Biol, 26:227–259, 1991.

[61] D.J. Wright, J.L. Rice, D.M. Yanker, and B.M. Znosko. Nearest neighbor param-
eters for inosine-uridine pairs in RNA duplexes. Biochemistry, 46:4625–4634,
2007.
[62] T. Xia, J. SantaLucia, M.E. Burkard, R. Kierzek, S.J. Schroeder, X. Jiao, C. Cox,
and D.H. Turner. Thermodynamics parameters for an expanded nearest-neighbor
model for formation of RNA duplexes with Watson-Crick base pairs. Biochem-
istry, 37:14719–14735, 1998.

5 See Also
New versions and related material can be found at
http://www.pasteur.fr/recherche/unites/neubiomol/meltinghome.html
https://sourceforge.net/projects/melting/
http://www.ebi.ac.uk/compneur-srv/melting/
You can use MELTING through a web server at http://bioweb.pasteur.fr/
seqanal/interfaces/melting.html http://www.ebi.ac.uk/compneur-srv/
melting/melt.php

73
6 Copyright
Melting is copyright c 1997, 2014 by Nicolas Le Novère, Marine Dumousseau and
William John Gowers.
This program is free software; you can redistribute it and/or modify it under the
terms of the GNU General Public License as published by the Free Software Foun-
dation; either version 2 of the License, or (at your option) any later version. This
program is distributed in the hope that it will be useful, but WITHOUT ANY WAR-
RANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more
details.
You should have received a copy of the GNU General Public License along with
this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place,
Suite 330, Boston, MA 02111-1307 USA

7 Acknowledgements
Thanks to Richard Owczarzy for reporting typos in several thermodynamic parameters
and reporting new public parameters. Nicolas Joly is an efficient and kind debugger
and advisor. Catherine Letondal wrote the HTML interface to melting. Thanks to
Nirav Merchant, Taejoon Kwon, Leo Schalkwyk, Mauro Petrillo, Andrew Thompson,
Wong Chee Hong, Ivano Zara for their bug fixes and comments. Thanks to Richard
Owczarzy for his magnesium correction. Thanks to Charles Plessy for the graphical
interface files. Thanks to Daniel McGreal for the SOAP interface of MELTING at
the EMBL-EBI. Thanks to Nicolas Rodriguez for the current web interface and finally
thanks to the usenet helpers, particularly Olivier Dehon and Nicolas Chuche.

8 Authors
Nicolas Le Novère, Marine Dumousseau and William John Gowers,
EMBL-EBI, Wellcome-Trust Genome Campus Hinxton Cambridge, CB10 1SD, UK
[email protected]

9 History
The Java version has been rewriten from the beginning. See the file ChangeLog for the
changes of the versions 4 and more recent.

74