bio-clinical-databases-somatic-signatures

name	bio-clinical-databases-somatic-signatures
description	Extract and analyze mutational signatures from somatic variants using SigProfiler or MutationalPatterns to characterize mutagenic processes. Use when identifying DNA damage mechanisms or etiology in cancer genomes.
tool_type	mixed
primary_tool	SigProfilerExtractor

Somatic Mutational Signatures

SigProfiler Workflow

Install and Generate Matrix

from SigProfilerMatrixGenerator import install as genInstall
from SigProfilerMatrixGenerator.scripts import SigProfilerMatrixGeneratorFunc as matGen

# Install reference genome (one-time)
genInstall.install('GRCh38')

# Generate mutational matrix from VCF
# Input: Directory containing VCF files
# Output: SBS96 matrix (96 trinucleotide contexts)
matrices = matGen.SigProfilerMatrixGeneratorFunc(
    project='my_project',
    genome='GRCh38',
    vcfFiles='/path/to/vcf_directory',
    plot=True,
    exome=False  # Set True for WES
)

Extract Signatures

from SigProfilerExtractor import sigpro as sig

# De novo signature extraction
# Determines optimal number of signatures automatically
sig.sigProfilerExtractor(
    input_type='matrix',
    output='extraction_output',
    input_data='my_project/output/SBS/my_project.SBS96.all',
    reference_genome='GRCh38',
    minimum_signatures=1,
    maximum_signatures=10,
    nmf_replicates=100,
    cpu=-1  # Use all cores
)

Decompose to COSMIC Signatures

from SigProfilerAssignment import Analyzer as Analyze

# Fit to known COSMIC signatures
Analyze.cosmic_fit(
    samples='my_project/output/SBS/my_project.SBS96.all',
    output='assignment_output',
    input_type='matrix',
    genome_build='GRCh38',
    signature_database='SBS_GRCh38_GRCh38'
)

MutationalPatterns (R)

Load and Analyze

library(MutationalPatterns)
library(BSgenome.Hsapiens.UCSC.hg38)

# Load VCF files
vcf_files <- list.files('vcf_dir', pattern = '\\.vcf$', full.names = TRUE)
sample_names <- gsub('.vcf', '', basename(vcf_files))

vcfs <- read_vcfs_as_granges(
    vcf_files,
    sample_names,
    ref_genome = 'BSgenome.Hsapiens.UCSC.hg38'
)

# Generate 96-context mutation matrix
mut_mat <- mut_matrix(vcf_list = vcfs, ref_genome = 'BSgenome.Hsapiens.UCSC.hg38')

# Visualize spectrum
plot_96_profile(mut_mat)

Fit to COSMIC Signatures

# Load COSMIC signatures (v3.2)
cosmic_sigs <- get_known_signatures(muttype = 'snv')

# Fit samples to signatures
fit_result <- fit_to_signatures(mut_mat, cosmic_sigs)

# Plot contribution
plot_contribution(fit_result$contribution, cosmic_sigs, mode = 'absolute')

# Relative contribution
plot_contribution(fit_result$contribution, cosmic_sigs, mode = 'relative')

De Novo Extraction

# Extract de novo signatures using NMF
# Determine optimal rank
estimate <- estimate_rank(mut_mat, rank_range = 2:8, nrun = 50)
plot(estimate)

# Extract signatures
nmf_res <- extract_signatures(mut_mat, rank = 4, nrun = 100)

# Compare to COSMIC
cos_sim <- cos_sim_matrix(nmf_res$signatures, cosmic_sigs)
plot_cosine_heatmap(cos_sim)

COSMIC Signature Etiology

# Common COSMIC signatures and their etiologies
SIGNATURE_ETIOLOGY = {
    'SBS1': 'Spontaneous deamination (age-related)',
    'SBS2': 'APOBEC activity',
    'SBS3': 'Defective HR/BRCA1/2',
    'SBS4': 'Tobacco smoking',
    'SBS5': 'Unknown (age-related)',
    'SBS6': 'MMR deficiency',
    'SBS7a': 'UV exposure',
    'SBS7b': 'UV exposure',
    'SBS10a': 'POLE mutation',
    'SBS10b': 'POLE mutation',
    'SBS13': 'APOBEC activity',
    'SBS15': 'MMR deficiency',
    'SBS17a': 'Unknown',
    'SBS17b': 'Unknown',
    'SBS18': 'ROS damage',
    'SBS22': 'Aristolochic acid',
    'SBS26': 'MMR deficiency',
    'SBS44': 'MMR deficiency',
}

def interpret_signatures(contributions):
    '''Interpret signature contributions'''
    interpretations = []
    for sig, contrib in contributions.items():
        if contrib > 0.05:  # >5% contribution threshold
            etiology = SIGNATURE_ETIOLOGY.get(sig, 'Unknown')
            interpretations.append({
                : sig,
                : contrib,
                : etiology
            })
     (interpretations, key= x: x[], reverse=)

Signature Categories

Category	Signatures	Mechanism
Age-related	SBS1, SBS5	Spontaneous deamination, clock-like
APOBEC	SBS2, SBS13	Cytidine deaminase activity
MMR deficiency	SBS6, SBS15, SBS26, SBS44	Mismatch repair defects
HR deficiency	SBS3	BRCA1/2, homologous recombination
POLE mutation	SBS10a, SBS10b	Proofreading defects
UV damage	SBS7a, SBS7b	Pyrimidine dimers
Smoking	SBS4	Tobacco carcinogens
Platinum therapy	SBS31, SBS35	Treatment-related

Cosine Similarity

import numpy as np

def cosine_similarity(sig1, sig2):
    '''Calculate cosine similarity between two signatures'''
    dot_product = np.dot(sig1, sig2)
    norm1 = np.linalg.norm(sig1)
    norm2 = np.linalg.norm(sig2)
    return dot_product / (norm1 * norm2)

# Threshold: >0.8 considered similar
# >0.9 considered same signature

Clinical Applications

def signature_clinical_implications(dominant_signatures):
    '''Clinical implications of mutational signatures'''
    implications = []

    for sig in dominant_signatures:
        if sig == 'SBS3':
            implications.append({
                'signature': 'SBS3',
                'implication': 'HR deficiency - may respond to PARP inhibitors',
                'testing': 'Consider BRCA1/2 testing'
            })
        elif sig in ['SBS6', 'SBS15', 'SBS26', 'SBS44']:
            implications.append({
                'signature': sig,
                'implication': 'MMR deficiency - may respond to immunotherapy',
                'testing': 'Consider MSI testing'
            })
        elif sig in ['SBS2', 'SBS13']:
            implications.append({
                'signature': sig,
                'implication': 'APOBEC activity - associated with high TMB',
                'testing': 'Consider TMB assessment'
            })

    return implications

Related Skills

• clinical-databases/tumor-mutational-burden - TMB calculation
• variant-calling/somatic-variant-calling - Input variants
• data-visualization/heatmaps-clustering - Signature visualization