Table 1 Methods for determining changes in the mutational spectrum
From: A Dirichlet-multinomial mixed model for determining differential abundance of mutational signatures
Methods for determining differential abundance in mutational signatures | |||
|---|---|---|---|
Name | Model | Caveats | Reference |
HiLDA | Signature extraction (LDA) followed by Dirichlet-multinomial | No mixed effects, local test not compositionally-minded | [23] |
Tumor Covariate Signature Model (TCSM) | Signature extraction and test through simulation | No mixed effects, test through simulation without interpretation of signatures in log-ratios | [24] |
Methods for studying the dynamics of mutational signatures | |||
|---|---|---|---|
TrackSig | Multinomial model to determine changepoints in bins of exposures | Compositional model but non-compositional interpretation of dynamics | [41] |
Clonesig | Clonal inference using potential differences in the exposures of clones as aid | Not a method for detecting differential abundance as such | [42] |
Log-odds of signature-specific exposures | Log-ratio of normalised exposures for each signature between early and late mutations, and between clonal and subclonal mutations | Non-compositional; changes in log-ratios of a signature can be due to changes in other signatures | [21] |