MMRD classification

MMRD classification

Using gradient boosting classifier combining microsatellite instability and mutational signatures

Currently available for tcga_mc3 and msk data settings. Example can be found in SigMA/examples/run_SigMA_MMRD.R

Step1: Create the input data table with 96 dimensional SBS counts and indel columns (nins, ndel, nmsi_del, nmsi_ins), for more info see the related page.

Step2: Run SigMA for signature estimations

output_SigMA <- run(genome_file = 'example_mmrd_SBS_matrix.csv',
                     data = 'tcga_mc3',
                     tumor_type = 'crc',
                     do_mva = F, # can also be set to T for Sig3 but not necessary for MMRD detection
                     check_msi = T,
                     catalog_name = 'cosmic_v3_inhouse')

Step2: Run classification

predict_mmrd(input_file = output_SigMA,
             data = 'tcga_mc3')

The above example is also in the examples directory:

cd SigMA/examples/
Rscript run_SigMA_MMRD.R

Based on signatures likelihoods only

Step1: Use run() function to calculate signatures

Step2: Use lite_df(), summarizes the findings and create general categories based on signature likelihoods.