Newbler 2

Roche Newbler 2.3 assembler was run using the command line (runAssembly) option with 90% sequence identity and 40-bp minimum overlap as parameters to perform de novo assembly of each of the Rhizobium genomes. Shell scripts were written to run runAssembly on multiple datasets. GSMapper 2.3 with 90% sequence identity and 40-bp minimum overlap was used to perform reference-based assembly of each genome using 305 core genes from strain 3841 as the reference genes. These 305 genes were those shown to be common to all chromid-bearing bacteria analysed by Harrison et al. [18 (link)]. Shell scripts were written to run runMapper on multiple datasets. Nucleotide information of 305 core genes was extracted from every draft genome using a Perl script, and a shell script was used to merge this information with their respective genes present in fully sequenced Rhizobium genomes. Each of the 305 files was aligned at nucleotide level by MUSCLE [22 (link)] that was run locally on the University of York Biology Linux grid. Each alignment file was checked and gaps were added for strains that had no reads for a given gene. The final results of FASTA alignments were concatenated by strain to form a 305-gene alignment using Galaxy [23 ]. A 100-gene alignment was also created using only those genes that were represented in every isolate by at least one read of at least 100 nucleotides.

The 454 reads were assembled using Newbler 2.3 (Roche) with default parameters. The Illumina assembly comprised of Illumina paired-end was obtained by using SOAPdenovo software^{48 (link)} (http://soap.genomics.org.cn) with default parameters. The contigs longer than 100 bp in the Illumina assembly were combined together with 454 reads for the final assembly using Newbler with default parameters.

FastQC [27 ] was applied for pre-trimming evaluation of the Fastq files containing the raw sequence reads. Trimming and primer removal were performed using cut-adapt and prinseq-lite [28 (link)]. Fastq files were error-corrected using RC454 [29 (link)] with the CSFV strain “Koslov” nucleotide sequence (GenBank HM237795; [26 (link)]) as reference. The de novo assembly was performed using Newbler 2.6 (Roche software, Basel, Switzerland). Each error-corrected Fastq file was aligned using the BWA-MEM [30 ] algorithm or Mosaik [31 (link)]. Subsequently, the libraries were post-processed with Samtools [32 (link)] and SNPs were called using lofreq [33 (link)]. Downstream SNP effect analysis was performed with snpEff [34 (link)] with no fraction cut-off set. SNP plots were made in R. dN/dS and πN/πS analyses were performed by taking the filtered SNP calls and running them through SNPGenie [35 (link)] with a SNP cut-off set at 1%. Statistical analysis was performed with Graphpad Prism using the T-test with Holm–Sidak correction for multiple comparisons when appropriate.

Library generation and genome sequencing of both strains were performed at “Darcy Fontoura de Almeida” Computational Genomics Unity (UGCDFA) of the National Laboratory of Scientific Computation (LNCC) (Petrópolis, Rio de Janeiro, Brazil). The genomes were sequenced using a whole-genome shotgun strategy, with a combination of Roche 454 GS-FLX shotgun and 3 kb-insert paired-end libraries. Libraries were prepared following GS FLX Titanium series protocols. 454 sequence reads were assembled using both Newbler 2.6 (454 Life Sciences, Roche Diagnostics Corporation, Branford, CT) and Celera (WGS, version 7.0) assemblers. The two assemblies were aligned to each other with Cross Match (Phred/Phrap/Consed package) since both yielded results that can be complementary and effective to close gaps. Sanger sequences previously obtained for strain CPAC 15 at Embrapa Soja (Brazil) [41 ] were also joined to the pyrosequence data. For closing gaps, a primer walking strategy was used, PCR products were Sanger sequenced at Embrapa Soja. All the consensus sequences of each contig and the sequence that closed each gap were aligned and joined using Consed (version 20.0). Information for scaffolding and gap closure was also obtained by mapping each assembly against the reference genomes of B. diazoefficiens USDA 110^T (BA000040.2) and B. japonicum USDA 6^T (AP012206.1).