Assembly of Phytoplasma Genome Drafts from Illumina Reads Using Phytoassembly

Abstract

Genome drafts for the phytoplasmas may be rapidly and efficiently assembled from NGS sequence data alone exploiting the proper bioinformatic tools and starting from properly collected samples. Here, we describe the use of the Phytoassembly pipeline (https://github.com/cpolano/phytoassembly), a fully automated tool that accepts as input row Illumina data from two samples (a phytoplasma infected sample and a healthy reference sample) to produce a phytoplasma genome draft, using the healthy plant host genome as a filter and profiting from the difference in reads coverage between the genome of the pathogen and that of the host. For phytoplasma infected samples containing >2% of pathogen DNA and an isogenic healthy reference sequence the resulting assemblies span the almost entire genomes.

Appendix: How Phytoassembly Works

• 1. The procedure uses the A5 pipeline to assemble the healthy plant sequence reads (Healthy.contigs.fasta), if no assembly was provided. The remaining files are archived. Next, the diseased plant reads are assembled (producing the file Diseased.contigs.fasta). A step in the A5 pipeline produces error corrected reads (Diseased.ec.fastq), which are used in all the subsequent steps.

• 2. The assembled reference sequence file is then indexed and aligned with the error corrected reads by the BWA tool [21] using the index and mem commands. Using the samtools (http://www.htslib.org/doc/samtools.html) commands view, sort, index and idxstats, a summary of statics is produced (Diseased.sorted.csv), consisting of the reference sequence name, sequence length, number of mapped, and unmapped reads.

• 3. This summary is passed to a phytocount.pl to estimate a cutoff value, by running once with cutoff 0, then using a fraction of the ratio between the sum of the lengths of the non-map** reads at cutoff 0 (Stage2.0.nonmatch.fastq, see below) and the sum of the lengths of the error corrected reads (Diseased.ec.fastq) of the diseased plant. Alternatively, if the user wants to supply a range of specifies fixed cutoff values, then the pipeline repeats the following steps from the lowest to the highest values provided (represented here as $cutoffval).

• 4. From the summary of statistical data (Diseased.sorted.csv), per-contig coverages are calculated and saved in a text file (Diseased.sorted.cov.csv).

• 5. The contigs with a coverage higher than $cutoffval are exported (Diseased.cutoff.$cutoffval.fasta, where $cutoffval is, e.g., “10”). The error-corrected reads from the diseased plant (Assembly.ec.fastq) are then aligned to the contigs in that last file using BWA (Stage1.$cutoffval.match.sam).

• 6. Using phytofilter.pl the reads above the cutoff from the alignment file are extracted and exported (Stage1.$cutoffval.match.fastq), using the sam flag #4 (“the query sequence itself is unmapped”) as filter.

• 7. These reads are now aligned with BWA against the healthy plant reference (Healthy.contigs.fasta), and the reads that do not align are exported (Stage2.$cutoffval.nonmatch.fastq). These non-aligned reads are assembled with the A5 pipeline (Stage3.$cutoffval.contigs.fasta).

• 8. A blast nucleotide database is created, using phytoblast.pl, from the reference healthy plant file (Healthy.contigs.fasta, which could also be a combination of different references) and used to query the contigs outputted by the previous stage (Stage3.$cutoffval.contigs.fasta) using tblastx. The results are saved in table (Stage3.$cutoffval.contigs.csv), which is then filtered according to the identity percentage (IP): entries with an IP greater than 95% are attributed to the plant (Stage3.$cutoffval.contigs.plant.csv), while those with a lower IP are attributed to the phytoplasma (Stage3.$cutoffval.contigs.phyto.csv). Using this last file the contigs pertaining to the phytoplasma are extracted from the query (Stage3.$cutoffval.phyto.fasta).

• 9. Lastly, the main outputs are archived and moved to a folder (Results_$timestamp), statistical data such as contigs size and number are calculated, and intermediate files are moved to a sub-folder (Other_files).