Program (and docker container instructions) for mapping CNV segment scores to a list of genes.
Input is an annotation in .bed format plus a directory that contains *.cnv files. This is the standard five column output for the Varscan CNV caller and consists of
sampleID chrom loc.start loc.end num.mark seg.mean
Output is a table with samples in columns and gene scores in rows. This file format is similar to gistic allgenes output.
To map segment scores to genes, we need to make decisions about duplicated genes and genes that span multiple segments.
For duplicated gene IDs in the input file, the code selects the most extreme score from the sample. This means that the score for one sample may come from a different genome location that the score for another. If you don't want the program to do this, please make sure your gene IDs are unique.
When a gene overlaps multiple segments, again the most extreme score (positive or negative) is reported.
Note: An earlier implementation of the code reported any such breakpoints but this does not fit in the gistic style format. If breakpoints are important to you, please contact the author.
To run cnvtogenes.py you can dowload it from the github repo directly:
wget https://raw.githubusercontent.com/jeltje/cnvtogenes/master/cnvToGenes.py
cnvtogenes.py requires the intervaltree package, which you can retrieve using pip install intervaltree.
Alternatively you can run the code via a docker container. For the examples below, I am assuming this.
The latest cnvtogenes docker image can be pulled directly from quay.io using
docker pull quay.io/jeltje/cnvtogenes
Alternatively, you can build from the github repo:
git clone https://github.com/jeltje/cnvtogenes.git
cd cnvtogenes
docker build -t jeltje/cnvtogenes .
For details on running docker containers in general, see the excellent tutorial at https://docs.docker.com/
To see a usage statement, run
docker run jeltje/cnvtogenes -h
docker run -v /path/to/input/files:/data jeltje/cnvtogenes --cnvdir <inputdir> --genefile <genome annotation.bed> > output.table
where
inputdir is the directory that contains .cnv files. You can get these through Varscan or ADTEx
genome annotation.bed is a bed format file containing your genes of interest.
It is important that the chromosome annotation (chr7 vs 7) matches between the bed file and the .cnv files. If your bed file uses chr but your cnv files do not, you can set the --nochr flag.
The program writes to STDOUT so you can capture the results using the > sign.
If you want to do a test run, please git clone https://github.com/jeltje/cnvtogenes.git and look in the test/ directory. Instructions are in the test/README.md file.
Important note: cnvtogenes.py does NOT apply any cutoffs to the scores.