Available online at www.sciencedirect.

com

Forensic Science International: Genetics 2 (2008) 184–189

www.elsevier.com/locate/fsig

A fast analysis system for forensic DNA reference samples

Johannes Hedman a,b,*, Linda Albinsson a, Carina Ansell a, Helene Tapper a,
Oskar Hansson a, Stig Holgersson a, Ricky Ansell a
a
Swedish National Laboratory of Forensic Science (SKL), SE-581 94, Linköping, Sweden
b
Department of Applied Microbiology, Lund University, Box 124, SE-221 00, Lund, Sweden
Received 19 October 2007; received in revised form 27 November 2007; accepted 17 December 2007

Abstract
On January 1st, 2006, the Swedish legislation on obtaining DNA reference samples from suspects and the recording of DNA profiles in databases
was changed. As a result the number of samples analysed at the Swedish National Laboratory of Forensic Science (SKL) increased from about 4500 in
2005 to more than 25,000 in 2006. To meet this challenge, SKL launched a new analysis system to create an unbroken chain, from sampling to
incorporation of a profile in the national DNA database and subsequent automatic generation of digitally signed hit reports. The system integrates
logistics, digital data transfer, new functions in LIMS (ForumDNA Version 4, Ida Infront AB) and laboratory automation. Buccal swab samples are
secured on a FTA1 card attached to an identity form, which is barcoded with a unique sample ID. After sampling, the police officer sends a digital
request to SKL. The sample is automatically registered in LIMS and processed on delivery. The resulting DNA profiles are automatically classified
according to quality using a custom-made expert system. Building the evaluation around mathematical rules makes it reproducible, standardised and
minimises manual work and clerk errors. All samples are run in duplicate and the two profiles are compared within LIMS before incorporation in the
database. In the first year of operation, the median time for completion of an analysis was 3 days, measured from delivery of the sample to
incorporation of the profile in the national DNA database. In spite of the dramatic increase in the number of reference samples there was no backlog.
# 2008 Elsevier Ireland Ltd. All rights reserved.

Keywords: Forensic DNA analysis; Reference sample; DNA database; Automation; LIMS

1. Introduction combine the automated analysis with improved laboratory

information management systems (LIMS) and other solutions
In the last decade, national DNA databases have been for digital data transfer [4,7]. The combination of automation
developed and continuously expanded throughout Europe [1]. and digitalised data handling has proved to be a powerful tool to
The legislations differ greatly between countries, resulting in speed up analysis, keep down the cost and avoid backlogs.
considerable differences in the number of database entries. In Sweden, the first legislation on DNA databases was passed
United Kingdom introduced a national DNA database in 1995 on April 1st, 1999 [8,9]. At that time, the entering of profiles to
[2]. In March 2006 it held over 3.7 million profiles from the national DNA database was restricted to crimes with possible
suspects and convicted felons, making it by far the largest in sentences above 2 years imprisonment. Up to 4500 reference
Europe [3]. In order to handle the large numbers of analyses, the samples were analysed annually at the Swedish National
Forensic Science Service developed an automated analysis Laboratory of Forensic Science (SKL). The samples were
process for buccal swabs [4]. Following legislative changes, predominately blood, analysed as single reactions and extracted
laboratories in other countries have developed their own using a manual, Chelex1 based method [10]. On January 1st,
automated systems, either based on cotton buccal swabs [5] or 2006, a new Swedish legislation on the acquisition of DNA
FTA1 cards [6]. Some laboratories have also made efforts to reference samples from suspects and the recording of DNA
profiles in databases was passed. Samples can now be taken
routinely from suspects, and the profile will be kept in the DNA
* Corresponding author at: Department of Applied Microbiology, Lund database if convicted to other sentences than fines [11,12].
University, Box 124, SE-221 00, Lund, Sweden. Tel.: +46 46 22 28 329;
fax: +46 46 22 24 203.
To prepare for the expected sample increase following the
E-mail address: [email protected] (J. Hedman). new legislation, SKL developed a new analysis system.

1872-4973/$ – see front matter # 2008 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.fsigen.2007.12.011
 J. Hedman et al. / Forensic Science International: Genetics 2 (2008) 184–189 185

Automated laboratory procedures are vital parts of this information management at the laboratory to sending profiles
system, but the greatest task was to establish ways of to the national DNA database and digitally signed hit reports
handling information digitally with high security and a back to the police. The system is supported by new logistics
minimum amount of manual paper work. Now all informa- solutions, an expert system for profile evaluation and new
tion is handled digitally throughout the analysis, from the functions in LIMS (ForumDNA, Version 4, IDA Infront)
web based request sent in by the police, via sample (Fig. 1).

Fig. 1. Overview of the flow of samples and digitalised information in the DNA reference sample analysis system. The police officer orders the analysis via a web
based digital request, monitors the analysis process via the web page and gets digital hit reports via e-mail. The samples are analysed and the profiles entered to the
national DNA database within a few days.
186 J. Hedman et al. / Forensic Science International: Genetics 2 (2008) 184–189

2. Materials and methods barcode of the plate and the sample ID:s on the identity forms,
and stores the information in a sample file together with the
2.1. Kit for collection of buccal swab samples sample positions. This file is exported to LIMS by clicking a
box, generating a complete batch for analysis.
A kit for buccal swab sampling was developed by SKL in Each batch holds 92 samples and two randomly distributed
collaboration with Nordkrim (http://www.nordkrim.se). The kit FTA1 controls (one positive and one negative). The punches
contains a FTA1 card (Indicating FTA1 Micro Card (Whatman are washed four times with de-ionised water using a SIAS
International Ltd.)) attached to an identity form, a Sterile Foam Xantus pipetting robot (Laboratory Automation & Technology
Tipped Applicator (Whatman International Ltd.), disposable A/S). Three complete plates can be washed simultaneously. To
gloves, a pre-addressed safety envelope with pre-paid postage, make the wash efficient, 150 ml of de-ionised water is
and written sampling instructions. A specific postal code is dispensed in three portions of 50 ml in each wash cycle,
used, hence the samples arrive to the laboratory sorted in followed by incubation for at least 5 min. The water is then
specific bags, separated from all other post. The identity form is removed and the discs are dried for 15 min using a miVac DNA
pre-barcoded with a unique sample ID, which follows the concentrator (Genevac Inc.).
sample from collection until a DNA profile has been obtained. The PCR setup is carried out on a SIAS Xantus robot, and
The police officer obtaining the sample records name, social the GeneAmp1 PCR System 9700 (Applied Biosystems) is
security number and status (i.e. suspect, victim or other) of the used with AmpF‘STR1 SGM Plus1 kit for amplification [14].
sampled person, his or her own name and police district, and a The PCR volume is set to 10 ml. In the first round of analysis 24
crime reference number. The physical link between the PCR cycles are run. The samples not generating acceptable
information-bearing identity form and the sample-bearing profiles are re-processed using 27 PCR cycles. The ABI 3130xl
FTA1 card minimises the risk of misplacing a sample during Genetic Analyzer (Applied Biosystems) is used for fragment
the process, and makes it easy for the laboratory staff to identify separation. The results are checked by one laboratory
and contact the responsible police officer if needed. technician using the GeneMapperTMID software, Version 3.1
(Applied Biosystems), with the analysis cut-off set to 50
2.2. Digital request, automated sample registration and relative fluorescence units (rfu). STR marker specific peak
information check height threshold values, ranging from 150 to 250 rfu for
heterozygote peaks and from 300 to 500 rfu for homozygote
ForumDNA LIMS is built on an iiPax 2 platform and uses an peaks, are used in the profile evaluation. Off-scale alleles are
object database [13]. The police officer orders the analysis of manually labelled or removed. Bleed-through peaks are
the reference sample electronically, using a digital request on a removed and comments are included for samples with too
restricted access web page. The information from the police is high levels of total DNA. The results are exported as a text file
automatically imported to LIMS and a reference sample case to a server, for automated classification.
with a unique case number is created. The analysis status of an
individual sample can be followed by the police via the web 2.4. Automated classification of DNA profiles
page. To facilitate comparison at the laboratory, the digital
request contains the same person and sample information as the A custom-made expert system based on Excel (Microsoft1
identity form. Excel 97 SR-2, Microsoft) and VBA-code (Microsoft Visual
When a reference sample reaches the laboratory, an Basic) is used for automated classification of the DNA profiles
information check is carried out within LIMS to ascertain according to quality. Each profile is given a quality class from 1
that the sample ID and social security number on the identity to 10, where 1 is a perfect profile and 9 is a blank. Class 10
form are identical to the corresponding information in the comprises profiles which need to be manually handled,
digital request. If there are no discrepancies, the sample is including mixtures or extra alleles possibly indicating a
processed immediately. LIMS puts any samples with mis- mutation, profiles with unbalanced peaks or potential drop out
matches onto an error list, which is handled by a laboratory alleles (Table 1).
technician. The text file from GeneMapperTMID is imported to the
expert system, which automatically generates: (i) a complete
2.3. Semi-automated DNA analysis list of samples and results, (ii) a classification of the profiles
into result classes, and (iii) a written report. The profile
Two punches from each FTA1 card are analysed in parallel, classification is based on four parameters: (1) maximum
on separate batches. A batch number is reserved in LIMS before allowed number of STR markers with unbalanced peaks
the run, and barcodes are printed. The LIMS batch is completed (defined as peak height ratio below 50%); (2) maximum
after punching, which takes away the need to sort the samples in allowed number of peaks for each STR marker; (3) number of
a specific order before the run. Punching of 1.2 mm FTA1 discs STR markers that must contain peaks; and (4) number of STR
into 96-well PCR plates is carried out on a BSD600 Duet semi- markers that must have peak heights above a defined percentage
automated punch robot (BSD Robotics). To avoid static of the threshold value (Table 1). The threshold values (peak
electricity, 80 ml of de-ionised water is added to each well height in rfu) are used as a measure of quality for each
before punching. The punch robot reads the batch number respective marker, and are updated after validation of each new
 J. Hedman et al. / Forensic Science International: Genetics 2 (2008) 184–189 187

Table 1 persons included in the database is kept in two separate police

Classification criteria for automated quality judgement of DNA profiles using a
databases, one for suspects and one for convicted felons. The
custom-made expert system
databases are connected to CODIS through LIMS, which also
Class A B C D holds information about the crime scene stains in the national
1 0 2 10 10 100% DNA database. The reference sample case is automatically
2 0 2 10 10 50% closed at the same time as the profile is transferred to CODIS.
3 0 2 10 7–10 50% Simultaneously, the profile will be searched against all profiles
4 0 2 10 0–6 50%
in the national DNA database, from both crime scenes and
5 0 2 5–9 1–9 50%
6 0 2 5–9 0 – persons. This process is run over-night. On the following day,
7 0 2 1–4 1–4 50% hit reports are automatically generated. The random match
8 0 2 1–4 0–4 0% probabilities are automatically calculated and turned into
9 0 0 0 0 – written conclusions following a general forensic scale in use at
10 10 6 10 10 0%
SKL. The hit reports are sent electronically to the police, using
A—Maximum allowed number of markers with unbalanced peaks. B—Max- digital signatures. Hits between persons and uncertain hits
imum allowed number of peaks for each marker. C—Number of markers that against stains are presented in a list for further investigation.
must contain peaks. D—Number of markers that must have peak heights above
a defined percentage of the threshold value. Class 1 includes complete profiles,
The entered profile remains in the national DNA database as
class 9 blanks and class 10 problematic profiles, e.g. mixtures. For a sample to long as the person remains in either the suspect or convicted
be automatically accepted by ForumDNA LIMS, the sum of the quality classes felon police database. Convicted felons usually remain in the
for two analyses must be five or less. A marker is considered to have unbalanced database for 10 years after the penalty has been served,
peaks if the height of the smaller peak is less than 50% of the height of the larger provided the person has not been involved in additional crimes.
peak. The threshold values (in rfu) are marker specific and are updated after
validation of each new batch of AmpF‘STR1 SGM Plus1 PCR mix. The
The profiles are removed automatically from the DNA database
thresholds typically range from 150 to 250 rfu for heterozygotes, and from 300 when a person is removed from the suspect or convicted felon
to 500 rfu for homozygotes. police database. Every night, the contents of the databases are
checked against each other, and profiles removed according to
batch of AmpFlSTR1 SGM Plus1 PCR mix. Profile the rules above.
classification is achieved through equations that set limits for
each of these four parameters. The limits can easily be changed 3. Results and discussion
without having to rewrite Excel equations or VBA-code. The
written report is saved as an Excel sheet on a server and The new DNA legislation has led to a significant increase in
exported to LIMS. the number of reference samples handled by SKL. The
LIMS uses the quality classification to decide if the profile digitalised and semi-automated analysis system has showed a
can be automatically accepted. If the profile is accepted, LIMS significant load capacity. It allows for high throughput handling
waits for the duplicate profile and compares the two profiles of DNA reference samples throughout the analysis and database
automatically. If no discrepancies are found, and the sum of the processes within just a few days. The whole system, excluding
result classes is five or less (e.g. one profile with class one and error investigation, is operated by approximately 1.5 full-time
the other class three), a so-called confirmed DNA profile is staff (60 man hours per week). About 270 DNA reference
created. The profiles from suspects are automatically exported samples can be handled each working day (540 single DNA
to the national DNA database during the following night. analyses). In the first year of operation the laboratory handled a
reference sample increase of more than 550%, from 4500 to
2.5. Investigating DNA results 25,000 samples, at the same time as duplicate analysis was
introduced, with no backlog. In 2007 the increase continued at a
LIMS transfers samples that are not accepted to error lists. If slower pace, reaching 28,500 analysed samples.
there is a mismatch in the duplicate, or the quality of the profiles Right after introduction, the median handling time was 5
is too poor, the laboratory technician manually evaluates the days, from delivery of a sample to entry of the profile to the
profiles, and decides if a confirmed DNA profile can be national DNA database. Hit reports are sent out digitally,
generated, or if the sample should be re-analysed. If the sample generally on the day after introduction in the database. After 6
does not produce a confirmed DNA profile after two rounds of months, the median handling time went down to 3 days, and it
punching, the sample is put through manual Chelex1 based has remained stable until January 2008. Within 5 days, over
DNA extraction [10]. 80% of the samples have been analysed, and the profiles entered
to the database. Within 2 weeks, 99% of the samples have been
2.6. Automatic transfer of DNA profiles to the national successfully handled. The troublesome samples include those
DNA database and creation of digital hit reports with low DNA amounts or PCR inhibition, and those with
incorrect sample information. Only a few samples with too little
Three data systems form the backbone of the Swedish DNA or too high levels of PCR inhibitors have had to be re-
national DNA database; CODIS, police databases and ordered from the police (24 out of over 25,000 during 2006).
ForumDNA LIMS. CODIS, operated by SKL, is used for Prior to the new legislation about 4500 DNA reference samples
storage and matching of profiles. The information about the were analysed in a single analysis process. The median
188 J. Hedman et al. / Forensic Science International: Genetics 2 (2008) 184–189

Table 2
Number of profiles in the Swedish national DNA database
January 2005 January 2006 January 2007 January 2008
Convicted felon profiles 3,455 4,276 8,549 16,796
Suspect profiles 1,331 1,839 15,183 24,621
Stain profiles 11,894 15,522 15,848 17,002
The database consists of three sections; profiles from suspects, profiles from convicted felons and profiles from crime scene stains. All DNA profiles from stains are
deleted from the database upon a complete profile hit against a person.

handling times then exceeded 2 weeks and multiple manual profiles is expected to approach 150,000. The suspect profiles
steps were required. increased more than eight-fold in 2006, but has seen a slower
At first, 27 PCR cycles were used for all samples. This growth in 2007. The number of profiles is expected to reach a
produced many profiles with bleed-through peaks that were steady state since persons only remain as suspects for a limited
difficult to evaluate. About 85% of the samples provide period of time before being removed from the database or kept
acceptable results for both punches in the first round of analysis, on as convicted felons (Table 2). The stain profiles are removed
performed with 24 PCR cycles. Re-punching and amplification after a complete profile hit against a person. Because of the
using 27 PCR cycles elevates the success rate, leaving about 2% increased number of persons to compare against, giving an
for manual DNA extraction using a Chelex1 based method elevated hit rate, the amount of stain profiles has almost
[10]. The second round of punching is always performed as a remained constant, after growing steadily the years before to
duplicate, for streamlining purposes, even if one of the discs the new legislation (Table 2). The number of hits between stains
gave a complete profile in the first round. There have been some and persons has increased significantly, whereas the number of
problems with static electricity in the punching of FTA1 discs, hits between different stains has decreased (Table 3).
especially during dry and cold weather. This has led to A new and standardised way of dealing with profile
‘‘jumping discs’’ or discs getting stuck in the punch evaluation has been introduced with the custom-made expert
mechanism. These discs are discarded in order not to system. Defining strict mathematical rules covering the aspects
compromise analysis security, and the samples are included otherwise controlled by a laboratory technician based on
in the 15% that need to be re-processed after the first analysis guidelines and experience, minimises manual work and clerk
round. Pre-filling the wells with water has diminished the errors. The advanced rules also enable efficient matching of
problem with static electricity, by binding the discs to the wells. partial profiles, thereby reducing the need to re-process
To improve the success rate further, the injection time on the samples. Since the rules are easy to change, the expert system
3130xl Genetic Analyzer was elevated from 5 to 10 s and an can be updated for changes in threshold values, new
optimised FTA1 disc wash protocol for the pipetting robot was amplification kit batches, or when changing or adding other
introduced. available STR amplification systems.
The main strategy for the analysis system is to let as many The strength of the reference sample analysis system
samples as possible pass through as quickly as possible, with a presented in this paper is the integration of logistics, data
strict batch handling. Rather than putting too much effort into transfer, custom-made expert systems and laboratory automa-
evaluating a questionable profile the sample is re-processed. tion to create a complete and unbroken chain, from sampling to
The strict rules as defined in Table 1, with threshold values well entering of the profile in the national DNA database and
above the cut-off limit of 50 rfu, ensures a high profile quality, automated generation of hit reports to be delivered electro-
but as a result a lower success rate than if more lenient rules are nically. Because of the limited need for manual input, the
used. The troublesome samples are generally re-processed throughput is easy to scale up within existing laboratory
within 2 days after delivery to the laboratory, giving a short premises, by primarily acquiring additional robots and
over-all handling time. instruments for PCR and capillary electrophoreses.
The Swedish national DNA database consists of three
sections; DNA profiles from suspects, convicted felons and Acknowledgement
crime scene stains, respectively. The number of convicted felon
profiles approximately doubled in both 2006 and 2007 The technical skill and enthusiasm of the DNA reference
(Table 2). Within 10 years, the number of convicted felon sample group at SKL is gratefully acknowledged.

Table 3
References
Number of hits between DNA profiles in the three sections of the Swedish
national DNA database [1] P.M. Schneider, P.D. Martin, Criminal DNA databases: the European
situation, Forensic Sci. Int. 119 (2001) 232–238.
2004 2005 2006 2007 [2] D.J. Werrett, The national DNA database, Forensic Sci. Int. 88 (1997) 33–
Stain–stain 2,750 3,271 2,897 1,418 42.
Stain–convicted felon 451 285 508 854 [3] The National DNA Database Annual Report 2005–2006, http://www.
Stain–suspect 553 857 5241 4,190 homeoffice.gov.uk/science-research/using-science/dna-database, United
Kingdom, 2006.
 J. Hedman et al. / Forensic Science International: Genetics 2 (2008) 184–189 189

[4] A. Hopwood, J. Brookes, A. Shariff, P. Cage, E. Tatum, R. Mirza, M. [9] Lagen om belastningsregister (Criminal Records Registry Act), SFS
Crook, K. Brews, K. Sullivan, A fully integrated robotic system for high 1998:620.
sample throughput within a DNA databasing unit, in: Proceedings from [10] P. Walsh, D. Metzger, R. Higuchi, Chelex 100 as a medium for simple
Promega User’s Symposium on Human Identification, vol. 8, 1997. extraction of DNA for PCR based typing from forensic material, Bio-
[5] M. Steinlechner, W. Parson, Automation and high throughput for a DNA techniques 10 (1991) 506–513.
database laboratory: development of a laboratory information manage- [11] Lag om ändring av Polisdatalagen (Amendment Act Concerning Police
ment system, Croat. Med. J. 43 (2001) 252–255. Data Act), SFS 2005:878.
[6] A.J. Hansen, B.T. Simonssen, C. Borsting, C. Hallenberg, N. Morling, [12] Lag om ändring av Rättegångsbalken (Amendment Act Concerning The
Semi-automated preparation of biological database samples for STR Swedish Code of Judicial Procedure), SFS 2005:877a.
typing, in: Int. Congr. Ser., vol. 1288, 2006, 663–665. [13] C. Karlsson, S. Holgersson, ForumDNA, a custom-designed laboratory
[7] W. Parson, M. Steinlechner, Efficient DNA database laboratory strategy information management system, in: Int. Congr. Ser., vol. 1239, 2003,
for high throughput STR typing of reference samples, Forensic Sci. Int. 783–786.
122 (2001) 1–6. [14] AmpF‘STR1 SGM Plus1 User’s guide, Applied Biosystems, Foster City,
[8] Polisdatalagen (Police Data Act), SFS 1998:622. USA.