demux_read_index

Demultiplex samples with in-read barcodes with spacers

Requirements

Install provided conda environment, otherwise requires

• python3
• HTSeq

Installation example

$ git clone https://github.com/cjowf/demux_read_index.git
$ cd demux_read_index
$ conda env create -f environment.yaml
$ conda activate demux_read_index
$ pip install .

Usage

$ demux_read_index -h
usage: demux_read_index [-h] -ss SS -r1 R1 -r2 R2 [-mm MM] [-fmmc] [-outdir OUTDIR]

This script performs demultiplexing of MiSeq paired-end reads

options:
  -h, --help      show this help message and exit
  -ss SS          SampleSheetFile Tab-delimited (headings:
                  Plate,Well,Index1,Spacer1,Index2,Spacer2,LabID,SampleID)
  -r1 R1          R1 fastq file
  -r2 R2          R2 fastq file
  -mm MM          Allow n barcode mismatches, default n=1
  -fmmc           Filter colliding barcode mismatches, otherwise disallow collisions
  -outdir OUTDIR  Output path, default current directory

Details

• See test/sample_sheet.tsv for example of sample sheet format
• All output files written to OUTDIR
• Each sample in sample sheet will have two files, both with indexes and spacers trimmed
  • <SampleID>_<Index1>_<Index2>_R1.fastq.gz
  • <SampleID>_<Index1>_<Index2>_R2.fastq.gz
• Reads with one or both barcodes not found in sample sheet are not trimmed, found in files
  • Undetermined_R1.fastq.gz
  • Undetermined_R2.fastq.gz
• Reads with both barcodes in sample sheet, but not in a combination found in the sample sheet will be trimmed, found in files
  • Undetermined_<Index1>_<Index2>_R1.fastq.gz
  • Undetermined_<Index1>_<Index2>_R2.fastq.gz
• Demultiplexed statistics written to OUTDIR/demux_stats.txt

Example workflow

1. bcl_convert for fastq generation. For the sample sheet it should be a single sample entry for all 16S samples, barcodes could be:
   • Dummy barcodes, eg AAAAAAAAA, desired reads in "Undetermined" fastq files.
   • Primers with index cycle primer hybridization sites will have a defined barcode sequence. Using these for bcl_convert can be a quality filter and/or could allow multiplexing with other types of samples. Desired reads in the sample_id fastq files.
2. demux_read_index.py for demultiplexing 16S samples.
   • All output files are primer and spacer trimmed, except Undetermined_R1 and Undetermined_R2.
   • Single threaded implementation, will take some time to run
   • No eta provided, will print '.' for every 10,000 reads
3. Check <out_dir>/demux_stats.txt for demuxed read counts. Note
   • Recognized barcodes in unrecognized combinations are put in Undetermined__...
   • For combinatorial barcodes: expect to see all possible combinations of forward and reverse barcodes, unrecognized combinations should have a small fraction of reads.

TODO

• Automate tests
• Add index cycle demultiplexing
• Performance improvements