-
Notifications
You must be signed in to change notification settings - Fork 1
Running
Andrew Roth edited this page Jan 4, 2017
·
2 revisions
THIS PAGE IS A WORK IN PROGRESS
The JointSnvMix software package consists of a number of tools for calling somatic mutations in tumour/normal paired NGS data.
After installation the jsm.py command should be available on your system.
There are two subcommands jsm.py classify and jsm.py train. If you type jsm.py classify -h or jsm.py train -h the list of options along with help will be presented.
There are currently three models supported by both the train and classify commands. All models use the JointSNVMix mixture model which jointly analyses the normal and tumour genomes.
By default snvmix2 is used but other models can be specified using the --model flag.
- binomial - Uses binomial densities in the mixture model this was previously referred to as the JointSnvMix1 mode.
- snvmix2 - Uses snvmix2 densities in the mixture as described in the original SNVMix paper previously referred to as JointSnvMix2.
- beta_binomial - Uses beta-binomial densities in the mixture model new in version 0.8. The beta-binomial is a robust (in the statistical sense) alternative to binomial model. It can be beneficial when dealing with over-dispersed data. This is useful in cancer genomes since allelic frequencies at somatic mutations sites may deviate significantly from those expected under diploid model.