-
Notifications
You must be signed in to change notification settings - Fork 3
Manual
pip3 install msmutect (not on pypi yet; this will not work yet)
Note: Cython must be preinstalled; if it is not, run: pip3 install Cython
git clone https://github.com/MaruvkaLab/MSMuTect_0.5
cd MSMuTect_0.5
sh rename.sh
pip3 install .
git clone https://github.com/MaruvkaLab/MSMuTect_0.5 for the first time; this downloads the code
On all runs do:
cd MSMuTect_0.5
export PYTHONPATH=$(pwd)
python3 src/Entry/main.py [flags]
If you are only interested in mutation loci, use the -m flag
If you are interested only in Alleles, use the -A flag
If you are interested only in Histograms, use the -H flags
If you are interested in all data, use all three, but bear in mind, this will make the program slower and the output files bigger
Sample run:
Python3 main.py -l [loci file] -T [tumor_file] -N [normal_file] -O [output prefix] -c [number of cores] -m
Histogram - a histogram is a structure containing indel information about a microsatellite locus. For instance, if the reference genome has motif repeats 5, but some cells have a motif-length deletion, a histogram could be: 5_10, 4_6. This means there were 10 reads with 5 motif repeats, and and 6 reads with 4 motif repeats
Alleles - Alleles are the alleles called by MSMuTect’s rigorous statistical noise model for each microsatellite locus. The output file contains the called alleles, their projected frequencies, and the extent to which the call precisely models the data (log likelihood). Mutations - Mutations are the mutations called by MSMuTect, by analyzing a normal and tumor sample from the same patient. It uses a rigorous noise model to weed out noisy loci, only analyzing loci that are good candidates for a mutation. The final step in the mutation calling process is a Fisher Exact Test for the histograms of the normal and tumor file for a locus. The p value of this test is also written to the output file (note that the majority of microsatellite loci will not have a Fisher Exact test performed since they do not pass the noise model).
The mutation calls are classified as follows:
-4: No normal alleles, or no tumor Alleles were able to be called (ie. at least one of the samples did not support alleles)
-3: Borderline non mutation
-2: Too many alleles were called. Probably very noisy region
-1: Insufficient read evidence to call alleles
0: no mutation
1: mutation
Run msmutect --help to see help message.
All flags
-T TUMOR_FILE, --tumor_file TUMOR_FILE
Tumor BAM file
-N NORMAL_FILE, --normal_file NORMAL_FILE
Non-tumor BAM file
-S SINGLE_FILE, --single_file SINGLE_FILE
Analyze a single file for histogram and/or alleles
-l LOCI_FILE, --loci_file LOCI_FILE
Loci to be processed and included in the output
-O OUTPUT_PREFIX, --output_prefix OUTPUT_PREFIX
prefix for all output files
-c CORES, --cores CORES
Number of cores to run MSMuTect on
-b BATCH_START, --batch_start BATCH_START
1-indexed number locus to begin analyzing at (Inclusive)
-e BATCH_END, --batch_end BATCH_END
1-indexed number locus to stop analyzing at (Inclusive)
-H, --histogram Output Histograms
-A, --allele Output alleles
-m, --mutation Output mutations
-F FLANKING, --flanking FLANKING
Length of flanking on both sides of an accepted read
-f, --force overwrite pre-existing result files, if they exist
MSMuTect can run for pairs of files (ie. a regular and a tumor), or call histograms/alleles for a single file. The -S flag denotes a single file, and -T and -N flags denote Tumor and Normal files, respectively. The -H flag will generate a histogram, the -A flag will generate allele calls, and the -m flag will generate mutation calls (for pairs of files only). MSMuTect will always show preceding steps. In other words, the -m flag will output a file with alleles and histograms in addition to the mutations, and the alleles flag will output a file with histograms as well. In addition, by default, MSMuTect will do more work, not less. So, if you give MSMuTect a single file, it will call alleles and histograms for it, and only by passing the -H flag (and not the -A flag) will it only call histograms.
Some examples:
$ msmutect -l generic_loci.phobos -S myBam.bam -O myout [-A]
Will generate a file containing the alleles and histograms for myBam.bam called myout.all.tsv. The -A argument is unnecessary, calling alleles is the default behavior.
$ msmutect -l generic_loci.phobos -S myBam.bam -H -O myout
Will generate a file holding only histograms called myout.hist.tsv. It’s first columns will be identical to myout.all.tsv
For pairs of files, by default MSMuTect will only analyze loci that are considered candidates for a mutation after analysis of the Normal file. So, for instance,
$ msmutect -l generic_loci.phobos -N normalBam.bam -T tumorBam.bam -O myout [-m]
Will output a file with mutations and some non mutations called myout.partial.mut.tsv
If you would like to see full data for a pair, for all loci, simply feed the -A or -H flags, along with the -m flag
$ msmutect -l generic_loci.phobos -N normalBam.bam -T tumorBam.bam -O myout -m -A
Will output a file myout.full.mut.tsv with all Alleles, Histograms, and Mutations for the Allele and tumor files
This will incur a significant performance penalty since all loci must be evaluated for both files
All files will record the mutation call for convenience
Please note, the -S and -T/-N flags cannot be used in tandem. -S is for single files. -T and -N is for running pairs
Sample locus data is available at https://drive.google.com/drive/folders/1DOioh0MNoH_rGmlat73BcH3aGqA72QyG?usp=sharing_eip_dm&ts=610a4722
for both HG37 and HG38.
Always pass the -c flag if you have more than one core. The runtime for a single core is very high
Count the mutations quickly and efficiently with cut -f49 [prefix].partial.mut.tsv | sort | uniq -c
Output prefix can include directories to easily place output files in separate directories
The Noise Table is written inline to ease distribution with pip. In any case, the hope is to replace this noise table with something more exact in the near future. The code is not completely pruned, however it is largely clear and concise. For now, the files are kept as .py to ease in debugging, which is the most important step in pre-1.0 versions. Many are changed by rename.sh before compilation. When the 1.0 version is created and uploaded to PYPI this will be changed
For support, please email [email protected] or [email protected]
The code is hosted at Github at https://github.com/MaruvkaLab/MSMuTect_0.5 . If there is some feature you would like added to make MSMuTect useful for you, please do not hesitate to email us and we will consider adding it. The code is licensed under an MIT license