A high-throughput pipeline for the design of real-time PCR signatures
© Satya et al; licensee BioMed Central Ltd. 2010
Received: 28 January 2010
Accepted: 23 June 2010
Published: 23 June 2010
Pathogen diagnostic assays based on polymerase chain reaction (PCR) technology provide high sensitivity and specificity. However, the design of these diagnostic assays is computationally intensive, requiring high-throughput methods to identify unique PCR signatures in the presence of an ever increasing availability of sequenced genomes.
We present the Tool for PCR Signature Identification (TOPSI), a high-performance computing pipeline for the design of PCR-based pathogen diagnostic assays. The TOPSI pipeline efficiently designs PCR signatures common to multiple bacterial genomes by obtaining the shared regions through pairwise alignments between the input genomes. TOPSI successfully designed PCR signatures common to 18 Staphylococcus aureus genomes in less than 14 hours using 98 cores on a high-performance computing system.
TOPSI is a computationally efficient, fully integrated tool for high-throughput design of PCR signatures common to multiple bacterial genomes. TOPSI is freely available for download at http://www.bhsai.org/downloads/topsi.tar.gz.
Rapid and accurate identification of pathogens from environmental and clinical samples is essential for effective containment of infectious diseases. Sequence-based identification methods, such as DNA microarrays and polymerase chain reaction (PCR) assays, are effective tools for pathogen diagnostics. Whereas PCR-based assays provide high specificity, microarray-based assays provide high multiplexing capability, accommodating thousands of oligonucleotide probes in a single diagnostic assay. The importance and utility of sequence-based identification methods have further increased in recent times due to advances in DNA sequencing technology that have led to the availability of a large number of pathogen genomes.
A variety of tools have been developed in the past few years to facilitate the design of pathogen-based diagnostic assays [1–10]. Notable among these are the whole-genome based signature design tools KPATH , Insignia [3, 11], and TOFI . Whereas KPATH designs signatures for PCR-based diagnostic assays and TOFI designs signatures for microarray-based assays, Insignia finds unique sequence segments that can be used to design both PCR and microarray signatures. While, among many features, these tools have the capability to identify common signatures shared by multiple target genomes, each has its own limitations. For example, KPATH computes consensus regions among the target genomes from their multiple alignments. However, multiple alignment of whole bacterial genomes is computationally intensive and it is not practical when a large number (> 20) of genomes is to be analyzed. Conversely, TOFI and Insignia build consensus regions among multiple genomes through pairwise alignments between the target genomes. Insignia server reports only the unique segments in the target genomes and provides an option for users to run the Primer3  PCR signature design software on these unique segments. Manual manipulation is necessary to extract the PCR signatures from Primer3 outputs and to perform further specificity analysis on the extracted signatures. Further manual manipulation is necessary when the unique segments reported by Insignia are not long enough to accommodate a complete PCR signature, in which case PCR signature components have to be designed individually from smaller unique segments close to each other, and the individual components have to be manually assembled to form valid PCR signatures. Insignia is extremely fast, as it precomputes the matches between all pairs of sequences. However, this advantage in speed comes with a limitation; the user is restricted to genomic sequences that are part of the Insignia database, and does not possess an option to use other sequences as targets or non-targets. The TOFI pipeline is free of most of the limitations described above, but can only design signatures for microarray-based diagnostic assays.
In this paper, we describe the Tool for PCR Signature Identification (TOPSI), which extends the TOFI framework [7–9] to design signatures for real-time PCR-based diagnostic assays. Like Insignia, TOPSI uses pairwise alignments to identify sequences that are common to multiple genomes, and compares these sequences with non-target genomes to identify unique segments suitable for designing signatures. However, TOPSI goes beyond the identification of unique segments, and incorporates modules to design PCR signatures from the unique segments and perform extensive specificity analysis on the designed signatures. Being fully integrated and automated, TOPSI takes a set of input target sequences and provides a list of PCR signatures common to all input targets without the need for manual intervention in any of the intermediate steps. Unlike existing software systems for real-time PCR signature design, TOPSI is the only one that is: freely available, high-throughput, and fully integrated. The following are some of the unique features of TOPSI:
Highly scalable: TOPSI is very efficient and scalable, as a result of using pairwise alignments as opposed to multiple genome alignments.
Fully integrated and automated: Complete PCR signature design and comprehensive specificity analysis are an integral part of TOPSI. PCR primers and probes are directly provided to the user, without the necessity for any manual manipulation.
Freely available: TOPSI is freely available for download and installation, giving users complete control over the selection of non-target databases and ensuring user confidentiality of the applications.
The TOPSI pipeline has been primarily designed to work with a large number of bacterial genomes. Designing common signatures for multiple viral genomes offers a different set of computational challenges. Although viral genomes are much smaller in size, signature design is complicated by the high variability within such genomes and the consequent lack of conserved regions suitable for signature design. As a result, the current TOPSI framework might not be successful in designing signatures common to a large number of viral genomes.
Overview of the pipeline
The input genomes are first compared with each other using the suffix-tree-based MUMmer  program in the pre-processing stage of TOPSI. Starting with an arbitrary pair of input genomes, pairwise local alignment is performed between the two genomes, and a list of conserved sequence segments that are shared between the two genomes is constructed from the pairwise alignment. This list of conserved sequence segments is then sequentially compared with each of the remaining input genomes and continually updated so as to contain only those sequence segments that are shared among all the input genomes. Designing PCR signatures from these conserved sequence segments ensures that each of the input genomes is amplified by the designed signatures.
The first stage of the core TOPSI pipeline uses the MUMmer program to perform pairwise comparison of the conserved target sequences with each non-target genome. This step eliminates any segments in the input sequences that have exact matches longer than a user-specified length with any of the non-target genomes in a comprehensive sequence database, such as the nt database provided by the National Center for Biotechnology Information (NCBI). The surviving segments, referred to as the candidate sequences, are then passed on to the second stage of the pipeline.
In the second stage, TOPSI uses the open source Primer3  software to identify primers and probes with the desired thermodynamic properties from the candidate sequences. At this stage of the pipeline, forward primers, reverse primers, and probes are designed independently, without taking into consideration the distance constraints between the primers and probes. This approach ensures that all unique primers and probes are reported, which can later be used to design PCR signatures in which only one, two or all three of the components are unique.
The third stage of TOPSI performs specificity analysis by performing BLAST  alignments of each primer and probe with any comprehensive sequence database provided by the user, such as the nt database. The BLAST alignments are performed in parallel on multiple processors using the blastn program of the parallel BLAST implementation mpiBLAST . Based on the BLAST alignments, primers and probes with significant alignments to non-target genomes are eliminated. The output of this stage consists of individual primers and probes that are unique to the target genome(s).
In the final post-processing step, the individual unique PCR primers and probes are assembled into PCR signatures by taking distance constraints into consideration. First, PCR signatures with all three unique components are identified. Further, PCR signatures with one or two unique components are also designed by taking each unique PCR primer or probe and designing non-unique components by running Primer3 on the conserved target sequence segments on either side of the unique components. These PCR signatures with one or two unique components are useful when there are very few or no PCR signatures with all three unique components.
Criteria for specificity analysis
TOPSI uses a combination of multiple criteria for performing specificity analysis, similar to TOFI . These criteria include the maximum percentage identity, the longest stretch of contiguous/near-contiguous matches, and the minimum number of mismatches with a non-target sequence. All these different criteria are evaluated based on the BLAST alignments obtained with non-target sequences. Whereas thresholds on the maximum percentage identity and the longest stretch of contiguous or near-contiguous matches are useful in evaluating the specificity of longer sequences, thresholds for the minimum number of mismatches with non-target sequences are useful in evaluating the specificity of short primer or probe sequences. Combining these criteria ensures that probes and primers of all different lengths are specific to the target genomes.
To enable the user to select a subset of the in silico designed PCR signatures, TOPSI assigns two different scores to each PCR signature. The first score, called uniqueness penalty, is a measure of the specificity of the PCR signature. The uniqueness penalty of each component of the PCR signature is calculated based on the best non-target match and the length of the longest contiguous match with a non-target sequence. A primer or probe with overall identity or the longest contiguous match exceeding pre-specified thresholds supplied by the user is assigned a penalty score of 1. Conversely, a primer or probe with no significant matches with non-target sequences is assigned a uniqueness penalty of zero. The uniqueness penalty for a PCR signature is computed as the sum of the uniqueness penalties of the individual components. The second score computed by TOPSI is the sum of the penalty scores reported by Primer3 for each of the three components, and is a measure of how close the thermodynamic properties of these components are to the optimal parameters selected by the user. The user can select a subset of the PCR signatures by ranking the probes based on any one of these scores, or by using a combined score calculated by assigning customized weights to each of the two scores.
In this section, we report signatures designed by TOPSI and compare them with those designed by other software systems as well as some experimentally verified signatures. We also discuss potential limitations of TOPSI.
Performance of the TOPSI pipeline
List of S. aureus genomes used for comparing TOPSI and KPATH
NCBI Taxon ID
S. aureus subsp. aureus Mu50
S. aureus subsp. aureus COL
S. aureus subsp. aureus JH1
S. aureus subsp. aureus JH9
S. aureus subsp. aureus MRSA252
S. aureus subsp. aureus MSSA476
S. aureus subsp. aureus Mu3
S. aureus subsp. aureus MW2
S. aureus subsp. aureus N315
S. aureus subsp. aureus NCTC 8325
S. aureus subsp. aureus str. Newman
S. aureus RF122
S. aureus subsp. aureus USA300 FPR3757
S. aureus subsp. aureus USA300 TCH1516
S. aureus EMRSA15 draft sequence
S. aureus 0582 draft sequence
S. aureus subsp. aureus str. JKD6008
S. aureus subsp. aureus str. JKD6009
Comparison with other software systems
Specificity thresholds used in TOPSI runs for S. aureus
M0 - longest stretch of contiguous matches with a non-target that has no mismatches
M1 - longest stretch of contiguous matches with a non-target that has at most one mismatch
M2 - longest stretch of contiguous matches with a non-target that has at most two mismatches
M3 - longest stretch of contiguous matches with a non-target that has at most three mismatches
Maximum overall identity with a non-target sequence
We attempted to obtain Insignia signatures using a total of 21 S. aureus genomes that were accessible through the Insignia Web server as of 4 June 2010, selecting the option to include NCBI RefSeq among non-targets and using a signature word length of 18 to match the corresponding parameter in TOPSI. Insignia produced a set of 68,879 signature chains (i.e., candidate regions for signature design) in less than a minute. As these signatures were too many for the Insignia Web server to run Primer3 or BLAST, we used the length filter to obtain 1,702 signature chains with length ≥ 28 bp. However, running Primer3 using the default parameters did not produce any PCR primers or probes, as the signature chains were not long enough (each was ≤ 51 bp) to accommodate all three PCR components. Individual PCR components could be designed independently, but Insignia does not provide any module to assemble adjacent PCR components into complete PCR signatures. Given that there were thousands of signature candidates, it was impractical to assemble the PCR components manually for the entire genome. To estimate the time necessary for specificity analysis, we used the BLAST search option in Insignia to submit 400 signature chains to the NCBI BLAST Web server, which took 1 hr and 26 minutes to produce the results. Assuming that the average time per query remains the same, it would take ~10 days to perform the BLAST analysis on the original 68,879 signature chains returned by Insignia. These results suggest that although Insignia might be extremely useful and convenient for designing a few signatures from selected regions of the target genome, unlike TOPSI, it is not ideal for high-throughput, whole-genome signature design on bacterial genomes that might result in thousands of signature candidates.
Comparison with experimentally verified signatures
To compare TOPSI signatures with experimentally verified signatures, we selected the extremely difficult case of PCR signatures that are unique to Burkholderia mallei with respect to Burkholderia pseudomallei. B. mallei and B. pseudomallei are closely related pathogens that cause different diseases, glanders and melioidosis, respectively . B. mallei is believed to have been clonally evolved from B. pseudomallei , with a significantly reduced genome due to the loss of genes. Due to the similarity of B. mallei with respect to B. pseudomallei, a literature search revealed only one PCR signature that was reported to be unique to B. mallei . However, both primers in this signature (5'-TTCGATCGATTCCTGCTATC-3' and 5'-GCGTTAAACGCCGTACTTTC-3') have exact matches with some newly sequenced B. pseudomallei strains. The Web-based tool Primer-BLAST http://www.ncbi.nlm.nih.gov/tools/primer-blast/index.cgi predicts that these primers will amplify B. pseudomallei strains 33, 172, 491, and 668. Hence, this PCR signature can no longer be considered to be unique to B. mallei.
Another set of 10 experimentally validated B. mallei specific PCR signatures were available from the Center for Bioinformatics and Computational Biology at the University of Maryland http://insignia.cbcb.umd.edu/pdf/burkholderia.pdf. These real-time PCR signatures were designed through their Insignia  system. However, BLAST comparisons of the primers and probes comprising these signatures with the NCBI whole-genome shotgun sequence (WGS) database in October 2009 revealed that five of these signatures did not meet our design criteria. Reasons for eliminating these signatures are listed in Additional file 1. The remaining five PCR signatures were still unique to B. mallei sequences.
List of 11 B. mallei genomes used for designing common signatures
NCBI Taxon ID
B. mallei ATCC 23344
B. mallei NCTC 10229
B. mallei NCTC 10247
B. mallei SAVP1
B. mallei PRL-20
B. mallei PRL7
B. mallei 2002721280
B. mallei ATCC 10399
B. mallei FMH
B. mallei GB8 horse 4
B. mallei JHU
TOPSI signatures for B. mallei.
Potential limitations of TOPSI
One potential limitation of TOPSI, and of other similar high-throughput signature design software systems, is their difficulty in designing signatures for viral genomes, as described by Philippy et al.. Because of their small genomes and high variability, it may not be possible to find conserved segments to design signatures from. We tested TOPSI with two viral agents, Variola major and human adenovirus. TOPSI identified six in silico PCR signatures (with at least one unique segment) common to 40 Variola major genomes, using three Variola minor genomes as non-targets, based on the classification provided by Esposito et al. . In contrast, TOPSI could not identify any common signatures for 16 human adenovirus genomes consisting of subgroups A, B, C, D, E and F. However, TOPSI could design unique PCR signatures common to two genomes of human adenovirus subgroup D. Our experience based on this limited testing with viral genomes indicates that it might be possible for TOPSI to design common signatures for large DNA viruses. However, TOPSI might not be able to design common signatures for short RNA viruses, in which case methods specifically designed for viral genomes, such as the one described by Duitama et al. , need to be incorporated.
Another issue of concern is the effect of draft or incomplete genomes on signature design. In the current TOPSI framework, PCR signatures are designed from genomic regions that are conserved among all the input genomes. This might potentially lead to a situation in which signatures common to a large number of input genomes are eliminated because of a single low-quality or incomplete genome sequence. One possible solution for this problem is to apply a lower threshold for consensus, so that signatures can be designed from regions that are conserved among a large percentage of the input genomes. This approach is compatible with the current TOPSI framework and could be incorporated into the system. However, this solution would lead to signatures that do not identify some of the target genomes. Therefore, it should be used only when signatures common to all targets cannot be identified. Another solution using the current TOPSI framework is to design signatures based solely on finished genomes as a first step, and subsequently filter the obtained signatures by applying a threshold on the percentage of draft (or incomplete) genomes that are identified by each signature. Alternatively, if sequence quality scores are available, taking them into consideration while evaluating the consensus regions might also lead to identifying signatures that might otherwise be eliminated.
The TOPSI pipeline is efficient in designing real-time PCR signatures that are common to multiple strains of a bacterial pathogen, and are also unique to the pathogen with respect to all other sequenced non-target genomes. Comparison with PCR signatures designed using a well-established software system shows that the TOPSI signatures are similar to those designed by the other software, and comparison with experimentally verified signatures shows that TOPSI is able to report signatures from unique regions of the pathogen genome. Being the only freely available, high-throughput, and fully integrated solution for the design of real-time PCR signatures, TOPSI provides a valuable contribution to the development of pathogen diagnostic assays.
Availability and requirements
Project name: TOPSI
Project home page: http://www.bhsai.org/downloads/topsi.tar.gz
Operating systems: Linux
Programming language: Perl
Other requirements: mpiBLAST 1.4.0 or later, MUMmer 3.19 or later, Primer3 1.1.4 or later, BioPerl, and a Linux cluster with PBS queuing system
TOPSI is also operational as a Web server at a U.S. Department of Defense (DoD) high-performance computing center. Sponsorship for access to these resources may be requested by contacting the corresponding author.
This work was sponsored by the U.S. DoD High Performance Computing Modernization Program, under the High Performance Computing Software Applications Institutes Initiative. We thank Mr. Tom Slezak of Lawrence Livermore National Laboratory for providing us with the PCR signatures designed by KPATH for S. aureus. We also thank Drs. David Kulesh and Leonard Wasieloski of the U.S. Army Medical Research Institute of Infectious Diseases for valuable discussions about the selection of parameters for PCR signature design.
The opinions and assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the U.S. Army or the U.S. Department of Defense. This paper has been approved for public release with unlimited distribution.
- Fitch JP, Gardner SN, Kuczmarski TA, Kurtz S, Myers R, Ott LL, Slezak TR, Vitalis EA, Zemla AT, McCready PM: Rapid Development of Nucleic Acid Diagnostics. Proceedings of the IEEE 2002, 90(11):1708–1720. 10.1109/JPROC.2002.804680View ArticleGoogle Scholar
- Kaderali L, Schliep A: Selecting signature oligonucleotides to identify organisms using DNA arrays. Bioinformatics 2002, 18(10):1340–1349. 10.1093/bioinformatics/18.10.1340View ArticlePubMedGoogle Scholar
- Phillippy AM, Mason JA, Ayanbule K, Sommer DD, Taviani E, Huq A, Colwell RR, Knight IT, Salzberg SL: Comprehensive DNA signature discovery and validation. PLoS Comput Biol 2007, 3(5):e98. 10.1371/journal.pcbi.0030098View ArticlePubMedPubMed CentralGoogle Scholar
- Rimour S, Hill D, Militon C, Peyret P: GoArrays: highly dynamic and efficient microarray probe design. Bioinformatics 2005, 21(7):1094–1103. 10.1093/bioinformatics/bti112View ArticlePubMedGoogle Scholar
- Rouillard JM, Herbert CJ, Zuker M: OligoArray: genome-scale oligonucleotide design for microarrays. Bioinformatics 2002, 18(3):486–487. 10.1093/bioinformatics/18.3.486View ArticlePubMedGoogle Scholar
- Slezak T, Kuczmarski T, Ott L, Torres C, Medeiros D, Smith J, Truitt B, Mulakken N, Lam M, Vitalis E, et al.: Comparative genomics tools applied to bioterrorism defence. Brief Bioinform 2003, 4(2):133–149. 10.1093/bib/4.2.133View ArticlePubMedGoogle Scholar
- Tembe W, Zavaljevski N, Bode E, Chase C, Geyer J, Wasieloski L, Benson G, Reifman J: Oligonucleotide fingerprint identification for microarray-based pathogen diagnostic assays. Bioinformatics 2007, 23(1):5–13. 10.1093/bioinformatics/btl549View ArticlePubMedGoogle Scholar
- Vijaya R, Zavaljevski N, Kumar K, Bode E, Padilla S, Wasieloski L, Geyer J, Reifman J: In silico microarray probe design for diagnosis of multiple pathogens. BMC Genomics 2008, 9(1):496. 10.1186/1471-2164-9-496View ArticleGoogle Scholar
- Vijaya R, Zavaljevski N, Kumar K, Reifman J: A high-throughput pipeline for designing microarray-based pathogen diagnostic assays. BMC Bioinformatics 2008, 9(1):185. 10.1186/1471-2105-9-185View ArticleGoogle Scholar
- Wang D, Urisman A, Liu YT, Springer M, Ksiazek TG, Erdman DD, Mardis ER, Hickenbotham M, Magrini V, Eldred J, et al.: Viral discovery and sequence recovery using DNA microarrays. PLoS Biol 2003, 1(2):E2. 10.1371/journal.pbio.0000002View ArticlePubMedPubMed CentralGoogle Scholar
- Phillippy AM, Ayanbule K, Edwards NJ, Salzberg SL: Insignia: a DNA signature search web server for diagnostic assay development. Nucleic Acids Res 2009, (37 Web Server):W229–234. 10.1093/nar/gkp286Google Scholar
- Rozen S, Skaletsky HJ: Primer3 on the WWW for general users and for biologist programmers. In Bioinformatics Methods and Protocols: Methods in Molecular Biology. Totowa, NJ: Humana Press; 2000:365–386.Google Scholar
- Kurtz S, Phillippy A, Delcher AL, Smoot M, Shumway M, Antonescu C, Salzberg SL: Versatile and open software for comparing large genomes. Genome Biol 2004, 5(2):R12. 10.1186/gb-2004-5-2-r12View ArticlePubMedPubMed CentralGoogle Scholar
- Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ: Basic local alignment search tool. J Mol Biol 1990, 215(3):403–410.View ArticlePubMedGoogle Scholar
- Darling A, Carey L, Feng W: The Design, Implementation, and Evaluation of mpiBLAST. In 4th International Conference on Linux Clusters: The HPC Revolution 2003 in conjunction with the ClusterWorld Conference & Expo. San Jose, CA; 2003.Google Scholar
- Ulrich RL, Ulrich MP, Schell MA, Kim HS, DeShazer D: Development of a polymerase chain reaction assay for the specific identification of Burkholderia mallei and differentiation from Burkholderia pseudomallei and other closely related Burkholderiaceae. Diagn Microbiol Infect Dis 2006, 55(1):37–45. 10.1016/j.diagmicrobio.2005.11.007View ArticlePubMedGoogle Scholar
- Godoy D, Randle G, Simpson AJ, Aanensen DM, Pitt TL, Kinoshita R, Spratt BG: Multilocus sequence typing and evolutionary relationships among the causative agents of melioidosis and glanders, Burkholderia pseudomallei and Burkholderia mallei. J Clin Microbiol 2003, 41(5):2068–2079. 10.1128/JCM.41.5.2068-2079.2003View ArticlePubMedPubMed CentralGoogle Scholar
- Esposito JJ, Sammons SA, Frace AM, Osborne JD, Olsen-Rasmussen M, Zhang M, Govil D, Damon IK, Kline R, Laker M, et al.: Genome sequence diversity and clues to the evolution of variola (smallpox) virus. Science 2006, 313(5788):807–812. 10.1126/science.1125134View ArticlePubMedGoogle Scholar
- Duitama J, Kumar DM, Hemphill E, Khan M, Mandoiu II, Nelson CE: PrimerHunter: a primer design tool for PCR-based virus subtype identification. Nucleic Acids Res 2009, 37(8):2483–2492. 10.1093/nar/gkp073View ArticlePubMedPubMed CentralGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.