Position weight matrix

This article is about Bioinformatics. For the disease

in horses known by the acronym PSSM, see Equine
polysaccharide storage myopathy.
A position weight matrix (PWM), also known as a

weights which could distinguish true binding sites from

other non-functional sites with similar sequences. Training the perceptron on both sets of sites resulted in a matrix and a threshold to distinguish between the two sets.[1]
Using the matrix to scan new sequences not included in
the training set showed that this method was both more
sensitive and precise than the best consensus sequence.[2]
The advantages of PWMs over consensus sequences have
made PWMs a popular method for representing patterns
in biological sequences and an essential component in
modern algorithms for motif discovery.[3][4]

PWMs are often represented graphically as sequence logos.

position-specic weight matrix (PSWM) or positionspecic scoring matrix (PSSM), is a commonly used 2 From Sequences to PWM
representation of motifs (patterns) in biological sequences.
A PWM has one row for each symbol of the alphabet:
PWMs are often derived from a set of aligned sequences 4 rows for nucleotides in DNA sequences or 20 rows
that are thought to be functionally related and have be- for amino acids in protein sequences. It also has one
come an important part of many software tools for com- column for each position in the pattern. In the rst
step in constructing a PWM, a basic position frequency
putational motif discovery.
matrix (PFM) is created by counting the occurrences
of each nucleotide at each position. From the PFM, a
position probability matrix (PPM) can now be created
1 Background
by dividing that former nucleotide count at each position
by the number of sequences, thereby normalising the
values. Formally, given a set X of N aligned sequences
of length l, the elements of the PPM M are calculated:

Mk,j =

N
1
I(Xi,j = k),
N i=1

where i (1,...,N), j (1,...,l), k is the set of symbols

in the alphabet and I(a=k) is an indicator function where
I(a=k) is 1 if a=k and 0 otherwise.
For example, given the following DNA sequences:
PWMs were introduced by American geneticist Gary Stormo.

The position weight matrix was introduced by American geneticist Gary Stormo and colleagues in 1982[1] as
an alternative to consensus sequences. Consensus se- the corresponding PFM is:
quences had previously been used to represent patterns
in biological sequences, but had diculties in the prediction of new occurrences of these patterns.[2] The rst

use of PWMs was in the discovery of RNA sites that

A 3 6 1 0
0
function as translation initiation sites. The perceptron alC
2
2
1
0
0

gorithm was suggested by Polish American mathemati- M = G1 1 7 10 0

cian Andrzej Ehrenfeucht in order to create a matrix of
T 4 1 1 0 10
1

6
2
1
1

7
1
1
1

2
1
5
2

1
2
.
1
6

4 INFORMATION CONTENT OF A PWM

The entries in the matrix make clear the advantage of adding pseudocounts, especially when using small
datasets

to construct M. The background model need not

A 0.3 0.6 0.1 0.0 0.0 0.6 0.7 0.2 have
0.1equal values for each symbol: for example, when
C 0.2 0.2 0.1 0.0 0.0 0.2 0.1 0.1 studying
0.2
. organisms with a high GC-content, the values
M=

G0.1 0.1 0.7 1.0 0.0 0.1 0.1 0.5 for0.1

C and G may be increased with a corresponding deT 0.4 0.1 0.1 0.0 1.0 0.1 0.1 0.2 crease
0.6 for the A and T values.

and therefore the resulting PPM is:

[5]

Both PPMs and PWMs assume statistical independence

between positions in the pattern, as the probabilities for
each position are calculated independently of other positions. From the denition above, it follows that the sum
of values for a particular position (that is, summing over
all symbols) is 1. Each column can therefore be regarded
as an independent multinomial distribution. This makes
it easy to calculate the probability of a sequence given
a PPM, by multiplying the relevant probabilities at each
position. For example, the probability of the sequence
S = GAGGTAAAC given the above PPM M can be
calculated:

When the PWM elements are calculated using log likelihoods, the score of a sequence can be calculated by
adding (rather than multiplying) the relevant values at
each position in the PWM. The sequence score gives an
indication of how dierent the sequence is from a random sequence. The score is 0 if the sequence has the
same probability of being a functional site and of being
a random site. The score is greater than 0 if it is more
likely to be a functional site than a random site, and less
than 0 if it is more likely to be a random site than a functional site.[5] The sequence score can also be interpreted
in a physical framework as the binding energy for that
sequence.

4 Information content of a PWM

p(S|M ) = 0.10.60.71.01.00.60.70.20.2 = 0.0007056.

Pseudocounts (or Laplace estimators) are often applied
when calculating PPMs if based on a small dataset, in order to avoid matrix entries having a value of 0.[6] This is
equivalent to multiplying each column of the PPM by a
Dirichlet distribution and allows the probability to be calculated for new sequences (that is, sequences which were
not part of the original dataset). In the example above,
without pseudocounts, any sequence which did not have
a G in the 4th position or a T in the 5th position would
have a probability of 0, regardless of the other positions.

The information content (IC) of a PWM is sometimes of

interest, as it says something about how dierent a given
PWM is from a uniform distribution.
The self-information of observing a particular symbol at
a particular position of the motif is:

log(pi,j )
The expected (average) self-information of a particular
element in the PWM is then:

Creating the PWM

pi,j log(pi,j )
Most often the elements in PWMs are calculated as log
likelihoods. That is, the elements of the PWM are transFinally, the IC of the PWM is then the sum of the exformed using a background model b so that:
pected self-information of every element:
Mk,j = log2 (Mk,j /bk ).
describes how an element in the PWM (left), Mk,j , can
be calculated. The simplest background model assumes
that each letter appears equally frequently in the dataset.
That is, the value of bk = 1/|k| for all symbols in the
alphabet (0.25 for nucleotides and 0.05 for amino acids).
Applying this transformation to the PPM M from above
(with no pseudocounts added) gives:

A 0.26
1.26
C
0.32
0.32
M=
G1.32 1.32
T 0.68 1.32

1.32
1.32
1.49
1.32

1.26
0.32
1.0 1.32
1.0 1.32

i,j

pi,j log(pi,j )

Often, it is more useful to calculate the information content with the background letter frequencies of the sequences you are studying rather than assuming equal
probabilities of each letter (e.g., the GC-content of DNA
of thermophilic bacteria range from 65.3 to 70.8,[7] thus
a motif of ATAT would contain much more information
than a motif of CCGG). The equation for information

content thus becomes

1.49 0.32 1.32
1.32 1.32 0.32
.
1.32
1.0
1.32

1.32
log(pi,j1.26
/pb )
0.32
i,j pi,j

3
where pb is the background frequency for that letter. This [9] Kel AE, et al. (2003). MATCHTM: a tool for
searching transcription factor binding sites in DNA secorresponds to the KullbackLeibler divergence or relaquences. Nucleic Acids Research. 31 (13): 3576
tive entropy. However, it has been shown that when using
3579. doi:10.1093/nar/gkg585. PMC 169193 . PMID
PSSM to search genomic sequences (see below) this uni12824369.
form correction can lead to overestimation of the importance of the dierent bases in a motif, due to the uneven
[10] Wrzodek, Clemens; Schrder, Adrian; Drger, Andreas;
distribution of n-mers in real genomes, leading to a sigWanke, Dierk; Berendzen, Kenneth W.; Kronfeld, Marnicantly larger number of false positives.[8]
cel; Harter, Klaus; Zell, Andreas (9 October 2009).

Using PWMs

ModuleMaster: A new tool to decipher transcriptional

regulatory networks. Biosystems. Ireland: Elsevier.
99 (1): 7981. doi:10.1016/j.biosystems.2009.09.005.
ISSN 0303-2647. PMID 19819296.

There are various algorithms to scan for hits of PWMs [11] Beckstette, M.; et al. (2006). Fast index based algorithms and software for matching position specic
in sequences. One example is the MATCH algorithm[9]
[10]
scoring matrices. BMC Bioinformatics. 7: 389.
which has been implemented in the ModuleMaster.
doi:10.1186/1471-2105-7-389.
PMC 1635428 . PMID
More sophisticated algorithms for fast database search16930469.
ing with nucleotide as well as amino acid PWMs/PSSMs
are implemented in the possumsearch software and are
described by Beckstette, et al. (2006).[11]

7 External links

References

[1] Stormo, Gary D.; Schneider, Thomas D.; Gold, Larry;

Ehrenfeucht, Andrzej (1982). Use of the 'Perceptron'
algorithm to distinguish translational initiation sites in
E. coli". Nucleic Acids Research. 10 (9): 29973011.
doi:10.1093/nar/10.9.2997.
[2] Stormo, G. D. (1 January 2000). DNA binding sites:
representation and discovery. Bioinformatics. 16
(1): 1623. doi:10.1093/bioinformatics/16.1.16. PMID
10812473.
[3] Sinha, S. (27 July 2006). On counting position weight
matrix matches in a sequence, with application to discriminative motif nding. Bioinformatics. 22 (14): e454
e463. doi:10.1093/bioinformatics/btl227.
[4] Xia, Xuhua (2012). Position Weight Matrix, Gibbs Sampler, and the Associated Signicance Tests in Motif Characterization and Prediction. Scientica. 2012: 115.
doi:10.6064/2012/917540.
[5] Guigo, Roderic. An Introduction to Position Specic
Scoring Matrices. http://bioinformatica.upf.edu. Retrieved 12 November 2013. External link in |work= (help)
[6] Nishida, K.; Frith, M. C.; Nakai, K. (23 December
2008). Pseudocounts for transcription factor binding
sites. Nucleic Acids Research. 37 (3): 939944.
doi:10.1093/nar/gkn1019.
[7] Aleksandrushkina NI, Egorova LA (1978). Nucleotide
makeup of the DNA of thermophilic bacteria of the
genus Thermus. Mikrobiologiia. 47 (2): 2502. PMID
661633.
[8] Erill I, O'Neill MC (2009). A reexamination of information theory-based methods for DNA-binding site identication. BMC Bioinformatics. 10: 57. doi:10.1186/14712105-10-57. PMC 2680408 . PMID 19210776.

3PFDB a database of Best Representative PSSM

Proles (BRPs) of Protein Families generated using
a novel data mining approach.
UGENE PSS matrices design, integrated interface to JASPAR, Uniprobe and SITECON
databases.

8 TEXT AND IMAGE SOURCES, CONTRIBUTORS, AND LICENSES

Text and image sources, contributors, and licenses

8.1

Text

Position weight matrix Source: https://en.wikipedia.org/wiki/Position_weight_matrix?oldid=736312107 Contributors: Michael Hardy,

BenFrantzDale, Ketil, Chowbok, Thorwald, Kbradnam, Rjwilmsi, SmackBot, Verne Equinox, RlyehRising, Colonies Chris, Sandve, Hsiaut,
CmdrObot, Montanabw, Opabinia regalis, IradBG, Addbot, DOI bot, Non-dropframe, Troels.M, , Gnomehacker, Citation bot 1,
DrilBot, Tom.Reding, Trappist the monk, RoadTrain, Amkilpatrick, RjwilmsiBot, Romeokienzler, Helpful Pixie Bot, Hsp90, Unknomics,
Fakharian.m, Monkbot, Harlba, Sbandya1 and Anonymous: 23

8.2

Images

File:Free-to-read_lock_75.svg Source: https://upload.wikimedia.org/wikipedia/commons/8/80/Free-to-read_lock_75.svg License: CC0

Contributors:
Adapted
from
<a
href='//en.wikipedia.org/wiki/File:Open_Access_logo_PLoS_white_green.svg'
class='image'
title='Open_Access_logo_PLoS_white_green.svg'><img
alt='Open_Access_logo_PLoS_white_green.svg'
src='//upload.wikimedia.
org/wikipedia/commons/thumb/9/90/Open_Access_logo_PLoS_white_green.svg/9px-Open_Access_logo_PLoS_white_green.svg.png'
width='9' height='14' srcset='//upload.wikimedia.org/wikipedia/commons/thumb/9/90/Open_Access_logo_PLoS_white_green.svg/
14px-Open_Access_logo_PLoS_white_green.svg.png 1.5x, //upload.wikimedia.org/wikipedia/commons/thumb/9/90/Open_Access_
logo_PLoS_white_green.svg/18px-Open_Access_logo_PLoS_white_green.svg.png 2x' data-le-width='640' data-le-height='1000'
/></a>
Original artist:
This version:Trappist_the_monk (talk) (Uploads)
File:ISMBECCB13-039.jpg Source: https://upload.wikimedia.org/wikipedia/commons/f/fd/ISMBECCB13-039.jpg License: CC BY
2.0 Contributors: Flickr: ISMBECCB13-039 Original artist: ismb
File:LexA_gram_positive_bacteria_sequence_logo.png Source: https://upload.wikimedia.org/wikipedia/commons/8/85/LexA_gram_
positive_bacteria_sequence_logo.png License: CC BY-SA 3.0 Contributors: Source: Created using the Weblogo software (http://weblogo.
berkeley.edu/), which is distributed on a MIT Open Source License (http://weblogo.berkeley.edu/LICENSE) Original artist: Gnomehacker

8.3

Content license

Creative Commons Attribution-Share Alike 3.0