bio-clinical-databases-somatic-signatures

What this skill does

## Version Compatibility

Reference examples tested with: SigProfilerMatrixGenerator 1.2+ (Bergstrom 2019), SigProfilerExtractor 1.1.24+ (Islam 2022), SigProfilerAssignment 0.1+ (Diaz-Gay 2023), MutationalPatterns 3.12+ (Manders 2022), MuSiCal 0.7+ (Liu 2024), SigNet (Serrano 2023, bioRxiv), HRDetect (Davies 2017 / Degasperi 2022 implementations), pandas 2.2+, R 4.3+. COSMIC v3.4 (September 2024) is the current reference catalog: 84 SBS, 11 DBS, 18 ID, 24 CN, 16 SV signatures.

Before using code patterns, verify installed versions match. If versions differ:
- Python: `pip show <package>` then `help(module.function)` to check signatures
- R: `packageVersion('<pkg>')` then `?function_name`

If code throws ImportError, AttributeError, or TypeError, introspect the installed package and adapt the example to match the actual API rather than retrying. COSMIC signature naming evolves: SBS40 was split to SBS40a/b/c in v3.4 (Senkin 2024); SBS17 split to SBS17a/b (5-FU); SBS10 split to SBS10a-d (POLE/POLD1).

# Somatic Mutational Signatures; Etiology, Extraction, Clinical Use

**'Extract mutational signatures from this tumor cohort and identify HRD/MMR/APOBEC processes'** -> Generate 96-context (or DBS/ID/CN/SV) matrix from VCF; choose de novo extraction (NMF) vs refit-to-COSMIC by cohort size; map dominant signatures to etiology; flag clinical actionability.

- Python (recommended): SigProfilerMatrixGenerator -> SigProfilerExtractor (de novo) or SigProfilerAssignment (refit)
- R alternative: `MutationalPatterns::fit_to_signatures()` (strict refit) or `extract_signatures()` (NMF de novo)
- Python (mvNMF for non-uniqueness): MuSiCal (Liu 2024 *Nat Genet*)
- Python (deep learning low-mutation count): SigNet (Serrano 2023)
- R (HRD-specific): HRDetect (Davies 2017 *Nat Med*); the 5-feature BRCA-deficiency classifier

## COSMIC v3.4 Catalog: Evolution and Composition

| Class | Count | Encoding |
|-------|-------|----------|
| **SBS** (Single Base Substitutions) | 84 | 96 trinucleotide contexts (6 substitution types x 16 trinucleotides) |
| **DBS** (Doublet Base Substitutions) | 11 | 78 strand-agnostic doublet classes (Bergstrom 2019) |
| **ID** (Insertion/Deletion) | 18 | 83 categories (indel length x repeat context x microhomology) |
| **CN** (Copy Number) | 24 | 48 channels (total CN x heterozygosity x segment length; Drews 2022 *Nature*) |
| **SV** (Structural Variants) | 16 | 32 channels (cluster x length x type) |

**Recent splits to know:**
- **SBS40 -> SBS40a / SBS40b / SBS40c** (Senkin 2024 *Nature*): pan-cancer active (40a); RCC-specific (40b/c).
- **SBS17 -> SBS17a (T>C uncertain) / SBS17b (T>G in CTT, 5-FU)** (Christensen 2019 *Nat Commun*; Pich 2019 *Nat Genet*).
- **SBS7 -> SBS7a / 7b / 7c / 7d** (UV photoproduct chemistry; Alexandrov 2020).
- **SBS10 -> SBS10a (POLE P286R) / 10b (POLE V411L) / 10c / 10d (POLD1)** (Mertz 2020 *Mol Cell*).

## Etiology Table (Postdoc-grade)

| Signature | Etiology | Clinical implication | Notes |
|-----------|----------|---------------------|-------|
| **SBS1** | Spontaneous 5mC deamination at CpG | Age-correlated; mitotic-rate biomarker | Clock-like |
| **SBS5** | UNKNOWN, clock-like; age-correlated | -- | Reviewer-accepted: "unknown, clock-like"; NOT polymerase fidelity errors |
| **SBS2 / SBS13** | APOBEC (APOBEC3A dominant per Petljak 2022) | Often co-occur; kataegis; ICI response signal | A3A vs A3B via YTCA vs RTCA tetranucleotide ratio |
| **SBS3** | HRD (BRCA1/2 deficient flat profile) | **PARP inhibitor eligibility** | HRDetect 98.7% sensitivity (Davies 2017) |
| **ID6** | HRD microhomology-mediated deletions | PARP eligibility | Pairs with SBS3 |
| **CN17 (HRD-CN1)** | HRD chromosomal instability | PARP eligibility; also BRCA1 promoter hypermethylation | Drews 2022 |
| **SBS6 / 14 / 15 / 20 / 21 / 26 / 44 + ID1 / 2** | MMR-D | ICI eligibility | Lynch typically 6/15/26/44; sporadic MLH1-hyperMet typically 21/26 |
| **SBS14 + SBS20** | POLE+MMR or POLD1+MMR double defect | Ultra-hypermutator; ICI excellent response | >500 mut/Mb |
| **SBS10a / 10b** | POLE-exo P286R / V411L | Hypermutator; ICI excellent response | 100-300 mut/Mb pure POLE |
| **SBS10c / 10d** | POLD1 | -- | Less common |
| **SBS28** | POLE indirect | Often co-extracted with SBS10 | -- |
| **SBS4 + DBS2** | Tobacco smoking; benzo[a]pyrene-G adducts | Lung cancer | C>A bias |
| **SBS7a/b/c/d + DBS1** | UV (CPD vs 6-4 photoproduct chemistry) | Melanoma | CC>TT dipyrimidine, CC>AA |
| **SBS24** | Aflatoxin | HCC (geographic) | C>A at CpC; Schulze 2015 *Nat Genet* |
| **SBS22** | Aristolochic acid | UTC, HCC | T>A at CpTpG; Hoang 2013 *Sci Transl Med* |
| **SBS17b** | 5-Fluorouracil | Therapy-induced | T>G in CTT context |
| **SBS31 / 35 / 86 / 87** | Platinum chemotherapy | Therapy-induced; second cancers | Cisplatin / carboplatin / oxaliplatin |
| **SBS11** | Temozolomide | Glioma post-TMZ | C>T at unmethylated CpC/CpT |
| **SBS88 + ID18** | Colibactin (pks+ E. coli) | CRC etiology; NTHL1-syndrome backgrounds | Pleguezuelos-Manzano 2020 *Nature* |
| **SBS30** | NTHL1 BER deficiency | Lynch-like; cancer predisposition | High cosine to FFPE artifact |
| **SBS-FFPE-artifact** | Formalin-induced C>T (NOT SBS33 as commonly cited) | Sequencing artifact | ~0.90 cosine to SBS30 (formalin-induced C>T characterization in mutational-signatures literature; specific paper attribution removed pending verification) |

**CRITICAL CORRECTION:** The widely-cited "SBS33 = FFPE artifact" is wrong. Modern literature attributes FFPE artifact to a signature resembling SBS30 (NTHL1-BER-deficiency profile); after enzymatic uracil repair the artifact instead resembles SBS1.

## Tool Taxonomy

| Tool | Approach | Class coverage | When to use | Fails when |
|------|----------|----------------|-------------|-----------|
| **SigProfilerSuite** (Alexandrov lab; Bergstrom 2019 *BMC Genomics*; Islam 2022 *Cell Genomics*; Diaz-Gay 2023 *Bioinformatics*) | Matrix gen -> NMF de novo / forward-backward refit | SBS / DBS / ID / CN / SV | Field standard; CPIC-equivalent for signatures | Heavy compute for de novo (100 NMF replicates) |
| **MutationalPatterns** (Manders 2022 *BMC Genomics*) | R-based; strict refit + NMF de novo | SBS / DBS / ID; lesion segregation | R workflows; reproducible refit | Lacks SV signatures |
| **MuSiCal** (Liu 2024 *Nat Genet*) | mvNMF (minimum-volume NMF) addressing NMF non-uniqueness | SBS / DBS / ID | Mid-size cohorts; novel signatures suspected | Less benchmarking at very large scale |
| **SigNet** (Serrano 2023, bioRxiv) | ANN-based signature attribution | SBS | Low mutation counts (best in 13-tool benchmark) | New tool; reproducibility data still maturing |
| **YAPSA** (Hubschmann 2020) | Linear combination decomposition | SBS | Comparison runs | Less widely used |
| **MutSignatures** (Fantini 2020) | Probabilistic refits | SBS | -- | -- |
| **deconstructSigs** (Rosenthal 2016) | NNLS (unregularized) | SBS | **DEPRECATED; never use** | NNLS overfits onto reference set; superseded by SigProfilerAssignment |
| **mSigAct** | Signatures from RNA-seq | SBS | RNA-seq only contexts | Limited resolution |
| **Helmsman** (Carlson 2018) | Fast matrix construction | SBS / DBS / ID | Preprocessing step only | Not for extraction/refit |
| **HRDetect** (Davies 2017 *Nat Med*) | Lasso logistic on 6 features (SBS3 / ID6 / RS3 / RS5 / HRD-LOH) | HRD-specific | BRCA1/2 deficiency classifier | Breast/ovarian-trained; cross-cancer needs revalidation |
| **MutationTimer** (Gerstung 2020 *Nature*) | Mutation timing relative to CN states | SBS | PCAWG-style evolution | Requires Battenberg/ASCAT CN; >=30x coverage |

**The deprecation:** deconstructSigs is the most-cited signature tool in publications but is **operationally deprecated**. NNLS without regularization overfits onto the ~70-signature reference; reviewers flag manuscripts using it without SigProfilerAssignment sensitivity. Replace with SigProfilerAssignment or MutationalPatterns strict refit.

## De Novo vs Ref