TBSS-Signature

This project studies whether mitochondrial mutation spectra vary along a positional proxy for the time that mtDNA remains single-stranded. The current taxonomic scope is Mammalia throughout the project. The supplied mutation tables retain other vertebrate classes as immutable source data, but every analysis filters to mammals before forming cohorts. The repository is organised as a sequence of self-contained analyses. Each numbered folder contains its own notebook or script, generated tables, figures, and a README that documents that stage.

Project structure

1init_data/
|-- data/                         source datasets
|-- Check192spec.ipynb            input quality-control notebook
`-- README.md
2species_intersection/
|-- data/                         intersection tables
|-- figures/                      intersection plots
|-- count_common_species.py       reusable intersection code
|-- CommonSpeciesStatistics.ipynb
`-- README.md
3compare_t_genes/
|-- data/                         matched cohorts and summaries
|-- figures/                      spectrum comparison plots
|-- mutation_comparison.py        reusable analysis and plotting functions
|-- CompareTGenes.ipynb
`-- README.md
4tsss_gradient/
|-- data/                         TSSS metadata and summaries
|-- figures/                      mutation-gradient plots
|-- TSSSMutationGradient.ipynb
`-- README.md
5context192/
|-- data/                         matched context spectra and summaries
|-- figures/                      192-component and context-gradient plots
|-- signatures/                   Python/R COSMIC signature assignment
|-- context192_analysis.py        reusable context-spectrum functions
|-- Context192MatchedGenes.ipynb
`-- README.md
6amino_acid_shift/
|-- data/midori/                  cached MIDORI2 GB272 FASTA archives
|-- data/derived_midori/          source manifest and final QC sequences
|-- figures/                      ratio distributions and paired effects
|-- midori_analysis.py            Mammalia discovery, pairing, and selection
|-- amino_acid_shift.py           translation-table and sequence-QC helpers
|-- run_analysis.py               MIDORI extraction/QC entry point
|-- AminoAcidShift.ipynb
`-- README.md

Analysis order

1. 1init_data documents and checks the supplied mutation-spectrum tables.
2. 2species_intersection counts mammalian species shared by every gene combination and records the exact members of each matched cohort.
3. 3compare_t_genes compares 12-component spectra for two to five genes using the same species for every selected gene. The current CO1/Cytb example contains 104 mammalian species and includes paired boxplots for C>T, G>A, and the C>T/G>A and A>G/T>C within-profile ratios, with pre-specified one-sided paired Wilcoxon tests.
4. 4tsss_gradient explores changes in selected mutation types along an explicit proxy for single-stranded duration. Its current three-gene cohort contains 52 mammalian species. It includes normalized weights, pre-normalization opportunity-adjusted burdens, and two complementary-substitution ratios calculated within each species.
5. 5context192 asks whether the flanking context of four transition types changes across the 52-species mammalian three-gene cohort. It keeps structural opportunity zeros separate from possible but unobserved contexts. Its signatures substage also fits COSMIC v3.3 SBS96 activities to strand-aware high, low, and high-minus-low transformations of each matched mean gene spectrum using both SigProfilerAssignment (Python) and mSigAct (R).
6. 6amino_acid_shift discovers Mammalia directly in MIDORI2 GB272 and obtains exact CDS/protein pairs for ten heavy-strand Major Arc genes, from COX1 to CytB, without using a MutSpec species list. Its strict ten-gene QC cohort contains 1,932 taxa. The notebook compares (Asn+Lys)/Gly and Pro/(Phe+Leu[TTA/TTG]), draws paired ten-gene boxplots, and displays all 45 forward gene-pair effects in compact matrices. It does not plot species counts or availability.

The main input for stages 2--4 is 1init_data/data/MutSpecVertebrates12.csv.gz; stage 5 uses the corresponding MutSpecVertebrates192.csv.gz table. Stage 6 is independent of those inputs and uses only its cached MIDORI2 sequence archives. Run stages 1--4 first, then either or both stage-5 branches as needed; stage 6 can be run independently. Stage 6 keeps only a source manifest and the final QC sequence table on disk; its derived analytical tables remain in notebook memory.

See the README inside each stage for its assumptions, commands, outputs, and interpretation limits.

The positional proxy used in stages 4 and 5 is calculated from fixed human rCRS coordinates and is not a species-specific measurement. Restricting the project to mammals removes between-class pooling but does not remove phylogenetic dependence, uneven sampling among mammalian clades, or coordinate-system effects.