This repository contains the source code of the our publication Sun, J. X., He, Y., Sanford, E., Montesion, M., Frampton, G. M., Vignot, S., ... & Lipson, D. (2018). A computational approach to distinguish somatic vs. germline origin of genomic alterations from deep sequencing of cancer specimens without a matched normal. PLoS computational biology, 14(2), e1005965.. The SGZ method is developed to allow researchers to distinguish somatic vs. germline alterations in cancer specimens without a matched normal in NGS data.
FMI SGZ method does not require installation. The core method is implemented in python script fmiSGZ.py. Simply provide required files while calling the method in python terminals.
After clone the repository or download and extract all files, run script 'run_test.py' to test if the scripts are working as expected. 'run_test.py' runs the basic SGZ method and FMI SGZ method on four test samples (provided). If the process succeeded, a message '-------Test succeeded.-------' should be printed on the standard output.
All the scripts are developed under Python 2.7.6.
fmiSGZ.py: the core SGZ method developed in Foundation Medicine Inc. to predict germline/somatic originsbasicSGZ.py: a basic method to predict germline/somatic (inspired by Jones, S., Anagnostou, V., Lytle, K., Parpart-Li, S., Nesselbush, M., Riley, D. R., ... & Galens, K. G. (2015). Personalized genomic analyses for cancer mutation discovery and interpretation. Science translational medicine, 7(283), 283ra53-283ra53.)run_test.py: wrapper script to run both method on four test samples
The scripts were last tested successfully in python 2.7.14 with the following module dependencies:
- os
- sys
- glob
- argparse (version 1.1)
- logging (version 0.5.1.2)
- csv (version 1.0)
- scipy (version 0.19.1)
- numpy (version 1.13.3)
- filecmp
A line in a CNA file represents a segment in the genome, and each column means:
- CHR: chromosome number of the segment
- segStart: start locus of the segment
- segEnd: end locus of the segment
- mafPred: predicted minor allele frequency of the segment
- CN: total copy number of this segment in the tumor cell predicted by a copy number algorithm
- segLR: median of log-ratio of all exon and snp targets in this segment
- segMAF: median of minor-allele-frequencies of snp targets in this segment
- numMAtumorPred: predicted copy number of minor allele of this segment
- numLRProbes: number of SNP targets and exon targets for estimating segLR of this segment
- numAFProbes: number of SNP targets for estimating segMAF of this segment
- purity: purity estimation of the sample
- baseLevel: baselevel of the sample, which equals to tumor_puritytumor_ploidy+2(1-tumor_purity). Basically using baseLevel and purity we can get ploidy estimation.
The input file cna_model_file should be a file that contains the copy number model of the specimen you are processing, and an example is under SGZ/test/test_samples/ sample1.cna_calls.txt. The SGZ algorithm should accept any cna_model_file provided by the user, as long as it is formatted like the sample1.cna_calls.txt.
Unfortunately we are not able to share the copy number modelling script publically, as it is part of our commercial product FoundationOne CDx Also, our copy number algorithm has many components that are customized for our assay ( baitset design, data format etc.), and is normally not compatible with external data.
A potential way to generate the CNA file is by using the ASCAT algorithm developed by Peter Van Loo et al. in Allele-specific copy number analysis of tumors, and the software (R package) can be found Here.
Please contact Yuting He yhe@foundationmedicine.com or James Sun jsun@foundationmedicine.com if you have any questions.
Please cite the paper Sun, J. X., He, Y., Sanford, E., Montesion, M., Frampton, G. M., Vignot, S., ... & Lipson, D. (2018). A computational approach to distinguish somatic vs. germline origin of genomic alterations from deep sequencing of cancer specimens without a matched normal. PLoS computational biology, 14(2), e1005965. if you use SGZ in your publication.
Last updated on July 22nd, 2019