Skip to content

Latest commit

 

History

History
73 lines (46 loc) · 2.7 KB

README.md

File metadata and controls

73 lines (46 loc) · 2.7 KB
Raw

Pipeline for analyzing PhIP-seq data

See Larman et. al. for more information.

The pipeline assumes that the initial data set is a single .fastq file with all the reads, including a possible barcode in the header of each read.

First start by splitting the data into smaller chunks:

fastq2parts.py -i in.fastq -o workdir/parts -p 2000000

Then align each read to the reference PhIP-seq library using bowtie (making sure to set the right queue):

bowtie_parts_with_LSF.py -i workdir/parts -o workdir/alns -x path/to/index_name.ebwt -l workdir/logs_aln -q short_serial

Then reads are reorganized according to barcode. The mapping file should be a tab-separated file with the barcode sequence as the first column and the sample name as the second column. The sample names should be something that's friendly as a UNIX filename:

parts2barcodes.py -i workdir/alns -o workdir/barcodes -m mapping.tsv

Now we must generate the counts and p-values. There are two ways to proceed:

  • Generate a single count file and a single p-value file, and have them all calculated at the same time.
  • Perform counts and p-values on each sample separately. Each p-value calculation gets dispatched as a separate job.

For the single-file method:

alns2counts.py -i workdir/barcodes -o workdir/counts.csv -r input_counts.csv
counts2pvals.py -i workdir/counts.csv -o workdir/pvals.csv

For the parallel method (make sure to set the queue):

alns2counts_separated.py -i workdir/barcodes -o workdir/counts -r input_counts.csv
counts2pvals_separated.py -i workdir/counts -o workdir/pvals -q short_serial -l logs_pvals
ls -l workdir/pvals/*.csv | awk '$5 == 0 {print $8}' | xargs rm -f  # remove empty pval files

# NOTE: ensure that all empty files in workdir/pvals have been deleted
merge_columns.py -f 2 -i workdir/pvals -o workdir/pvals.csv

Note that any of these commands can be dispatched to the LSF job scheduler.

PGM inference model

There is a Gibbs sampling method for computing "selection values" as well. This script expects a CSV file with three columns: the clone names, the input counts, and the output counts.

gibbs.py --input counts.csv --output output.csv --verbose

The --verbose flag will instead create a directory called output and also dump a bunch of diagnostic figures there.

There will soon be an mcmc.py script that will implement the more complex PGM that allows multiple timepoints.

Other things to note

If your CSV file is polluted with Windows-style line-endings, you can change that with:

tr '\r' '\n' < offending.csv > clean.csv