Skip to content

milaboratory/mixcr-rna-seq-paper

Repository files navigation

Docker image for MiXCR performance analysis with in silico generated data

The Docker image contains pre-installed software and scripts to generate in silico RNA-Seq data and run performance analysis.

Running analysis

Docker image is already built and deposited in Docker Hub. To run the full analysis just install Docker and execute the following line (no any special downloads required, Docker will automatically download the image):

docker run -v WORKING_PATH:/work -i milaboratory/rna-seq-paper:v1.0 /bin/bash -c "/opt/scripts/run-comparison.sh && /opt/scripts/run-false-positives.sh" > log 2>errLog

where WORKING_PATH should be replaced with some particular path at your host computer. Be sure that you have at least ~80Gb of RAM and ~1TB of disk space.

Description

Scripts run-comparison.sh and run-false-positives.sh will execute the following steps:

  • generate a set of TRB VDJ recombinations with repseqio tool
  • generate fastq files with in silico RNA-seq data with a) a portion of TRB records b) without any TCR records
  • run MiXCR analysis in RNA-Seq mode
  • run STAR aligner to produce BAMs and then run TRUST
  • compare results obtained by MiXCR and TRUST with the original set of CDR3 clones and draw comparison plots
  • calculate rate of false-extensions produced by MiXCR extendAlignments action
  • calculate rate of false-overlaps produced by MiXCR assemblePartial action

All intermediate data can be found in the WORKING_PATH.

The following resulting files will be produced:

  • SupplementaryFigure_in_silico.pdf — comparison of MiXCR and TRUST results
  • falseExtensionsResults.txt — MiXCR false-positive extension rate estimations
  • falseOverlapsResults.txt — MiXCR false-positive partial assembly rate estimations

For the details on available options see run-comparison.sh script.