Mutational Signature Enrichment (MSE) Calculator

This is the code for the Mutation Signature Enrichment calculating tool used in these publications:

Boichard, Pham, et. al APOBEC-related mutagenesis and neo-peptide hydrophobicity: implications for response to immunotherapy
Pham, Boichard, Goodman, et. al Role of Ultra-Violet Mutational Signature versus Tumor Mutation Burden in Predicting Response to Immunotherapy (manuscript)

It is UCSD CCAL's implementation of the PMAC-D metric from Roberts, et. al An APOBEC Cytidine Deaminase Mutagenesis Pattern is Widespread in Human Cancers, called AMSE (APOBEC Mutational Signature Enrichment) in the APOBEC paper above, used to measure the degree of APOBEC mutagenesis from a list of mutations called in a sample. PMAC-D counted the following nucleotide changes and their reverse complements, weighted equally:

TCA ==> TGA
TCA ==> TTA
TCT ==> TGT
TCT ==> TTT

In addition to the PMAC-D signature definition included by default, it has also been generalized to calculate the MSE statistic based on probabilities for each of the 96 trinucleotide switches as defined by Alexandrov, et. al and available on the COSMIC Mutational Signatures page. These mutation signature probabilities are defined in the tab-delimited file /data/signatures_probabilities.txt.

Usage

The code accepts called mutations in MAF files, with the same column order as found on TCGA, or VCF files as inputs.

from compute_mutational_signature_enrichment import compute_mutational_signature_enrichment

mutation_file_paths = ["./TCGA_test/TCGA-44-7661.maf","./TCGA_test/TCGA-CQ-7068.maf","./TCGA_test/TCGA-DD-A73G","./TCGA_test/TCGA-FF-A7CX.maf"]

compute_mutational_signature_enrichment(
    mutation_file_paths, #list of called mutation file paths in MAF or VCF format, 1 file per sample
    reference_file_path, #reference genome FASTA, must be the same as the one used to call the mutations in the mutation_file_paths
	cosmic_mutation_signature_number=0 #Default 0 uses the defined PMAC-D weights, any number from 1-30, inclusive corresponding to the Alexandrov signatures can be used. 
	)

The output is a pandas dataframe where the columns are sample file names and the rows are information related to the calculation for a particular sample. The signature MSE score is in the row "Mutational Signature Enrichment".

This code should work on any computer that can run Python 3.

Dependencies:

pandas
pyfaidx (don't forget to index reference FASTA file)
pprint

Example output of 5 TCGA tumors using APOBEC Signatures PMAC-D/AMSE, Signature 2, and Signature 13:

Sample	AMSE	Signature 2	Signature 13
TCGA-44-7661.maf	2.5153333995826297	1.696616476895444	2.122606877977259
TCGA-CQ-7068.maf	2.5070760447035956	1.7943464758950072	2.628330781579929
TCGA-DD-A73G.maf	0.9325034978530419	0.7367417842643569	1.1181360888995775
TCGA-FF-A7CX.maf	0.6527777777777777	0.7307526870927789	0.5969015874220198

The ENSEMBL GRCh37 full genome DNA reference (ftp://ftp.ensembl.org/pub/grch37/current/fasta/homo_sapiens/dna/Homo_sapiens.GRCh37.dna.primary_assembly.fa.gz), which was used to call the TCGA mutations, was used. Please see the TCGA_test folder for the exact TCGA MAF files used to generate this table. The script TCGA_test.py can be used to generate the above table.

Comparison with PMAC-D on Pan-Cancer TCGA Data:

Info