C-Sibelia: an easy-to-use and highly accurate tool for bacterial genome comparison
At 2013-11-25 09:58:11 AM | 423
The development of inexpensive genome sequencing technologies and efficient assembly methods has revolutionized the study of bacterial genomes, which are being sequenced and assembled on a daily basis. When an assembly is available, the most common first task is to compare it against a reference genome (or another assembly, if no such genome is available) in order to find genetic differences between the newly assembled and reference genomes. This analysis is critical to understand genetic factors that determine certain phenotypes of the isolates.
We present Comparative Sibelia software (C-Sibelia) for the comparison of two bacterial genomes in the form of complete sequences or draft assemblies. C-Sibelia is able to compare genomes in the presence of rearrangements and duplications. C-Sibelia takes as input two FASTA files (the assembly and reference files; if the reference genome is not available, it can be substituted by another draft assembly) and produces: (1) a VCF file containing all identified single nucleotide variations (SNVs) and indels; (2) annotation of these variants by SnpEff; (3) an XMFA1 file containing alignment information. The web-based version also produces a circular diagram visualizing the rearrangement of synteny blocks in two genomes.
The performance of C-Sibelia in detecting SNVs and indels is comparable to MUMmer and outperforms Mauve in terms of the false-positive rate. C-Sibelia is a part of the Sibelia comparative genomics suite, which is freely available under the GNU GPL v.2 license at https://sourceforge.net/projects/sibelia-bio. Users are encouraged to use the web-based version of C-Sibelia at https://etool.me/software/csibelia.
From synteny blocks to alignment
The task of finding SNVs and indels connects closely to the problem of whole-genome alignment. Unlike aligning two short DNA segments, aligning two genomes is more challenging because of the presence of rearrangements and repetitive elements. C-Sibelia addresses this problem by first decomposing genomes into synteny blocks, using theiterative de Bruijn graph algorithm described in Minkin et al.2. This step separates linear operations (indels, substitutions) from non-linear operations (rearrangements) and thus allows us to apply global alignment to multiple instances of each synteny block. C-Sibelia incorporates LAGAN3, a global alignment tool, for aligning different instances of the same synteny block.
From alignment to variant calling. C-Sibelia then finds differences between two genomes (indels, SNVs, rearrangements) by analyzing the resulting synteny and alignment blocks. Regions in one genome not covered by synteny blocks are treated as indels. SNVs and small indels that lie within the regions covered by synteny blocks are reported by analyzing the alignment information produced by LAGAN. Identified variants are annotated by using snpEff4. The pipeline of C-Sibelia is described in the following pseudocode.
Input: An assembly and a reference genome (in FASTA format).
Decompose the sequences into synteny blocks using Sibelia.
Align instances of synteny blocks using LAGAN.
Analyze the synteny block decomposition and alignment information.
– Find indels in non-syntenic regions.
– Find small indels and SNVs in aligned regions (using the alignment information produced by LAGAN).
– Annotate the identified variants using SnpEff.
– Select contigs containing multiple synteny blocks (i.e., rearranged contigs).
All SNVs and indel variants, in a VCF file.
Annotation of these variants produced by SnpEff4.
A picture in Circos format5 for rearranged contigs and the reference genome.
A simulated dataset
From the complete genome of Staphylococcus aureus (S. aureus) NCTC 8325, we performed 10 deletions of random segments of size 2000 bp, and futher introduced 1000 SNVs in the resulting genome. We then generated five reversals and five translocations of random segments in the genome with size 10,000 bp each to evaluate the capability of these tools to perform an alignment in the presence of rearrangements. We obtained a simulated assembly of this newlysimulated genome of 180 contigs; the distribution of contig length was similar to that of the RN4220 assembly reported in Dhanalakshmi et al.8. We further used C-Sibelia, Mauve and MUMmer to find variants in this simulated assembly and the original reference genome (NCTC 8325). Table 1 and Table 2 demonstrate that the performance of C-Sibelia in detecting variants is comparable to MUMmer and improves upon Mauve in terms of the false-positive rate. Figure 1shows the Circos diagram of the rearranged contigs and the reference genome. The scripts and commands used for this benchmark are available in the Supplementary material.
Table 1. SNV calling on simulated data.
Table 2. Indel calling on simulated data.
Figure 1. A picture in Circos format for assembly sequences and the reference genome.
Only contigs with multiple synteny blocks rearranged differently in the genome are shown. Green and red bars depict the direction of synteny blocks on the positive and negative strands, respectively.
A real dataset
The most common approach for comparing an assembly against a reference genome is to first align the assembly against the reference and then write in-house scripts to extract variants. C-Sibelia can achieve this task automatically and with high accuracy. We used C-Sibelia to reproduce the comparison of the S. aureus RN4220 assembly and the reference genome NCTC 8325, reported in Dhanalakshmi et al.8 (the authors used MUMmer and in-house scripts for this comparison). Among 132 single nucleotide variants and four large deletions reported in Dhanalakshmi et al.8, C-Sibelia confirmed 121 SNVs and all four large deletions. C-Sibelia also reported six additional variants, which are also confirmed by BLAST9. The input data as well as the commands for generating these results are available in theSupplementary material.
The Etool Web-Server
The online version of C-Sibelia is available at https://etool.me/software/csibelia. The web form takes as input two FASTA files (one for the assembly and the other for the reference). The web form’s parameters allow users to choose whether or not to annotate variants and display the Circos5 picture for rearrangement analysis (see Figure 1). Results are delivered to registered users by a real time push notification mechanism10,11.
In this application note, we introduced C-Sibelia, a novel software for comparing two closely-related bacterial strains. Performance of C-Sibelia is comparable to MUMmer, and better than Mauve in terms of false positives rate. The web interface of C-Sibelia makes the task of comparing assemblies against a reference genome convenient for microbiologists, who do not want to go to the trouble of downloading and compiling the software. In the future, we plan to extend C-Sibelia to compare multiple genomes or draft assemblies as well as scale the software to larger genomes.
SP designed the project. IM, HP, ES, NV performed experiments and developed the computational tool. IM, SP wrote the manuscripts with critical input from NV, ES, HP. All authors read and approved the final manuscript.
No competing interests were disclosed.
This work was partially supported by Russian Foundation for Basic Research RFBR (grant 14-04-31926) and V-P foundation.The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
We would like to thank Alla Lapidus, Ekaterina Chernyayeva, Ryan Tewhey, Xiao Li, Harish Nagarajan, Mikhail Kolmogorov, Dhanalakshmi Nair, and David Hernandez for many helpful discussions. We are indebted to Phillip Compeau for for many helpful editorial suggestions.
- 1. Brudno M, Poliakov A, Salamov A, et al.: Automated whole-genome multiple alignment of rat mouse, and human. Genome Res. 2004; 14(4): 685–692. PubMed Abstract | Publisher Full Text | Free Full Text
- 2. Minkin I, Patel A, Kolmogorov M, et al.: A scalable and comprehensive synteny block generation tool for closely related microbial genomes. arXiv preprint arXiv: 1307.7941, 2013. Publisher Full Text
- 3. Brudno M, Do CB, Cooper GM, et al.: LAGAN and Multi-LAGAN: efficient tools for large-scale multiple alignment of genomic DNA. Genome Res. 2003; 13(4): 721–731. PubMed Abstract | Publisher Full Text |Free Full Text
- 4. Cingolani P, Platts A, Coon M, et al.: A program for annotating and predicting the effects of single nucleotide polymorphisms, SnpEff: SNPs in the genome of Drosophila melanogaster strain w1118; iso-2; iso-3. Fly (Austin). 2012; 6(2): 80–92. PubMed Abstract | Publisher Full Text | Free Full Text
- 5. Krzywinski M, Schein J, Birol I, et al.: Circos: an information aesthetic for comparative genomics. Genome Res. 2009; 19(9): 1639–1645. PubMed Abstract | Publisher Full Text | Free Full Text
- 6. Darling AE, Mau B, Perna NT: progressiveMauve: multiple genome alignment with gene gain, loss and rearrangement. PloS One. 2010; 5(6): e11147. PubMed Abstract | Publisher Full Text | Free Full Text
- 7. Kurtz S, Phillippy A, Delcher AL, et al.: Versatile and open software for comparing large genomes. Genome Biol. 2004; 5(2): R12. PubMed Abstract | Publisher Full Text | Free Full Text
- 8. Nair D, Memmi G, Hernandez D, et al.: Wholegenome sequencing of Staphylococcus aureus strain rn4220, a key laboratory strain used in virulence research, identifies mutations that affect not only virulence factors but also the fitness of the strain. J Bacteriol. 2011; 193(9): 2332–2335. PubMed Abstract | Publisher Full Text| Free Full Text
- 9. Tatusova TA, Madden TL: BLAST 2 sequences, a new tool for comparing protein and nucleotide sequences.FEMS Microbiol Lett. 1999; 174(2): 247–250. PubMed Abstract | Publisher Full Text
- 10. Brandt S, Kristensen A: Web push as an internet notification service. In W3C Workshop on Push Technology, Boston, Massachusetts, 1997. Reference Source
- 11. Fette I, Melnikov A: The websocket protocol. IETF Internet draft. 2011. Reference Source