PlasmidSeeker

A k-mer based program for the identification of known plasmids from bacterial whole genome sequencing reads

PlasmidSeeker has been published in PeerJ and can be accessed from here

=========================

USING PLASMIDSEEKER TOOL

1. INSTALLING AND BUILDING DATABASE

• Make sure you have PERL and R installed
• Put GenomeTester 4 binaries (gdistribution, glistcompare, glistquery, glistmaker) to a directory named "GenomeTester4", which should be directly under the main directory, which contains Testfunction.R, plasmidseeker.pl and database_builder.pl
• Put all plasmid FASTA files together into a single multi-fasta file (using UNIX cat command, for example; multi-fasta file with 8,514 plasmids: http://bioinfo.ut.ee/plasmidseeker/plasmid_db_12jul17.fna.gz)
• Use database_builder.pl to create the database or download our database with 8,514 plasmids (k=20) from our department server (http://bioinfo.ut.ee/plasmidseeker/) or from FigShare: https://figshare.com/s/5f7b924544839f7d6e59
• Approximate time with 8,514 Refseq plasmids, k=20 with 32 cores and 512GB RAM was 11 minutes.
• For simplified installation and testing, follow the readme under "example" directory.

command line example: perl database_builder.pl -i [multi-FASTA file with all plasmids]

Database builder options:

• -i - Input fasta file with all plasmid sequences
• -d - Database directory (default „plasmid_db“, will be created if does not exist)
• -t - Number of threads used (default 32)
• -w - K-mer length used (default 20)

2. DETECTING PLASMIDS

• Download closest bacterial reference genome to your isolate of interest (one possible source is NCBI Refseq: https://www.ncbi.nlm.nih.gov/refseq/)
• Use plasmidseeker.pl to detect plasmids from your isolate samples
• Approximate time using the 8,514 Refseq plasmids database, k=20 with 32 cores and 512GB RAM and 140 Mbp sample is less than 3 minutes.
• command line example: perl plasmidseeker.pl -d [your database dir] -i [your WGS sample FASTQ file] -b [close reference bacterium FASTA file] -o [output file name]

PlasmidSeeker options (printed with no input):

• -i - Input FASTQ file location
• -o - Output file name (default is "plasmidseeker_result.txt")
• -d - Path to plasmid database directory
• -b - Closest reference bacterium FASTA file location
• -t - Number of threads used (default 32)
• -f - Minimum threshold F - at least this fraction of unique k-mers that has to be found for a plasmid (default 80)
• -c Percent used to cluster plasmids
• -k Keep temporary plasmid distribution files, save plasmid distribution graphs, save additional summary
• -a Coverage variation - how much coverage variation (0-100%) is allowed (due to normal differences in sequencing coverage, default 0%). Could be relevant for larger genomes, when bacterial and plasmid sequences have markedly different composition or sequencing is biased.
• -h Print help
• --verbose Print out more working process
• --ponly Assumes that reads contain only plasmid sequences (use for extracted plasmids)

3. EXAMINING OUTPUT

• Reference plasmids which share more than 80% of k-mers are presented in a single cluster (example "CLUSTER 1") and ordered by the percentage of unique k-mers found. Likely, only one plasmid of each cluster is present in the sample.
• "HIGH P-VALUE PLASMIDS" refer to plasmids whose copy numbers are similar to the isolate. These plasmids may be integrated or false positives.
• Output example: "E_coli_wgs.txt" (E. coli WGS sample); "P_aeruginosa_plasmid.txt" (P. aeruginosa plasmid)
• The flag "-k" saves individual plasmid distribution files and saves a png graph of the bacterial (blue) and plasmid (red) distributions. In addition the fitted curves are shown and thresholds in between which the analysis is done. Additional .txt file shows various information about the plasmid. For example, how many of the reference's k-mers are predicted to be in the sample for both bacteria and plasmid with the multiplicated proportion also shown.