bio-comparative-genomics-whole-genome-duplication

What this skill does

## Version Compatibility

Reference examples tested with: wgd v2.0.31+ (heche-psb/wgd; Chen & Zwaenepoel 2024 Bioinformatics 40:btae272), KsRates 1.1.3+ (VIB-PSB/ksrates; Sensalari 2022 Bioinformatics 38:530), DupGen_finder (Qiao 2019 Genome Biol 20:38), MAPS 1.0 (Li 2018), POInT (Conant lab), SLEDGe (bioRxiv 2024.01.17.574559), Whale.jl 2.0+, ksrates pip 1.1+, MCScanX 1.0+, PAML 4.10+ (yn00/codeml for Ks), BLAT 36+, DIAMOND 2.1+, R 4.4+, mclust 6.1+ (for mixture models). Python 3.10+ required for wgd v2.

Before using code patterns, verify installed versions match. If versions differ:
- CLI: `wgd --version`; `ksrates --version`; `wgd ksd --help`
- Python: `pip show wgd ksrates`
- R: `packageVersion('mclust')`

If code throws `wgd ksd: cannot find PAML output`, `KsRates: insufficient sister species`, `MAPS: missing tree`, these tools have specific input expectations: wgd needs codon-aware MAFFT/MUSCLE alignment; KsRates needs configured `config_ksrates.txt`; MAPS needs nucleotide tree. The deprecated arzwa/wgd v1 is replaced by heche-psb/wgd v2.

# Whole Genome Duplication Analysis

**"Are there WGD events in this lineage and when did they occur?"** -> WGD detection combines **Ks distributions** (synonymous-substitution rates between gene paralog pairs, showing peaks at past polyploidy events) and **synteny block analysis** (parallel collinear blocks within a genome). Modern best practice uses **wgd v2** (Chen & Zwaenepoel 2024 Bioinformatics 40:btae272) as an integrated pipeline. **KsRates** (Sensalari 2022 Bioinformatics 38:530) is mandatory for cross-lineage comparison because substitution rates vary across the tree -- ignoring this places WGDs incorrectly relative to speciation events. The fundamental tradeoff: Ks plot peaks are visually obvious but biologically ambiguous between (1) small-scale tandem duplications, (2) segmental duplications, and (3) true WGD; combining Ks with synteny anchors disambiguates these.

- CLI: `wgd dmd` and `wgd ksd` -- paranome construction + Ks distribution
- CLI: `wgd syn` -- synteny-anchored WGD signal extraction
- CLI: `ksrates init` then `ksrates wgd-paralogs ortho` -- substitution-rate-corrected positioning
- CLI: `MCScanX -h` then `dupgen_finder` -- duplication-class assignment
- CLI: `mapsR` -- gene-tree-based WGD phylogenetic placement

## Algorithmic Taxonomy

| Tool | Approach | Output | Strength | Fails when |
|------|----------|--------|----------|------------|
| wgd v2 (Chen & Zwaenepoel 2024 Bioinformatics 40:btae272) | Integrated paranome + Ks + synteny + WGD dating | Ks distributions, collinearity plots, GMM/ELMM mixture fits, dating | Standard 2024 pipeline; replaces deprecated arzwa/wgd v1 | Saturation at Ks > 2; single-lineage substitution rate variation |
| KsRates (Sensalari 2022 Bioinformatics 38:530) | Substitution-rate correction via outgroup pairs | Adjusted Ks ages of focal-species paralogs vs orthologs | MANDATORY when comparing WGDs across lineages with different rates | Requires at least 2 outgroups for rate calibration |
| DupGen_finder (Qiao 2019 Genome Biol 20:38) | Classifies duplications by genomic context | Per-gene class: tandem, proximal, dispersed, segmental, WGD | Disambiguates duplication type | Class assignment depends on intervening-gene-count windows |
| MAPS (Li 2018) | Phylogenomic placement of WGD via gene-tree topology mapping | WGD position on species tree | Detects WGDs from gene-tree-species-tree discordance | Computationally heavy; requires many gene trees |
| POInT (Conant 2008 Genome Res 18:1597) | Order-aware reconstruction of WGD chromosomes | Reconstructed ancestral WGD genome | Strong inference for syntenic-block ages | Lineage-specific tuning required |
| SLEDGe (bioRxiv 2024.01.17.574559) | ML classifier on Ks plot features | WGD vs no-WGD binary call + confidence | Reduces visual-peak-fitting subjectivity | Newer; less validated |
| wgd v1 (arzwa/wgd; DEPRECATED) | Predecessor of v2 | -- | Historical | Use v2 (heche-psb/wgd) |
| dupHMM (Zwaenepoel 2019 different MBE paper) | HMM on Ks for WGD/SSD attribution | Per-pair WGD/SSD classification | Probabilistic disambiguation | Less integrated than wgd v2 |
| Whale.jl (Zwaenepoel & Van de Peer 2019 MBE 36:1384) | Bayesian DTL + WGD reconciliation | WGD posterior at species-tree nodes | Native WGD modeling; integrates with [[gene-tree-species-tree-reconciliation]] | Julia ecosystem |
| WGDexploreR (legacy) | Visualize Ks plots | Plots only | Visualization aid | Not for inference |
| McLuster-WGD (custom workflows) | Mixture-model fitting on Ks | GMM / ELMM components | For custom peak fitting | Not a standard tool |
| ksrates web (sensalari 2022) | Web interface to ksrates | Same as CLI | User-friendly | Manual config; not scriptable for genome-wide |
| FastKaKs (Kang 2017) | Fast Ks computation for many pairs | Ks values per pair | Faster than PAML yn00 | Slightly less accurate; for screening |
| wgs2pep (Vandepoele/wgs) | Pep-to-WGD ortholog identification | Per-paralog WGD assignment | Plant-focused | Less general |

Methodology evolves; verify the current wgd v2 manual and the Cheng & Zwaenepoel 2024 review chapter (in *Polyploidy: Methods and Protocols*) before locking on a single workflow. The 2R vertebrate / 3R teleost / Ss4R salmonid WGDs are well-established; novel WGD claims require concordance across Ks, synteny, and phylogenomic placement.

## Decision Tree by Experimental Scenario

| Scenario | Recommended approach | Why |
|----------|------------------------|-----|
| Plant comparative genomics; suspect WGD | wgd v2 + KsRates | Modern standard pipeline; plants are WGD-prone |
| Vertebrate 2R or 3R WGD analysis | wgd v2 + MAPS phylogenomic placement | Vertebrate-specific; combines Ks + tree topology |
| Salmonid Ss4R WGD | Whale.jl with WGD node in species tree | Native WGD modeling in Bayesian framework |
| Distinguish WGD from sequential small duplications | Ks GMM fit + synteny block analysis (wgd syn) | Ks peak alone insufficient; synteny confirms |
| Date WGD relative to a speciation event | KsRates with outgroup speciation calibration | Required when substitution rates differ |
| Identify duplicates retained vs lost after WGD | DupGen_finder + Ks distribution | Classification + age estimation |
| Detect novel WGD in non-model organism | wgd dmd -> wgd ksd -> visual GMM/ELMM mixture | Standard discovery workflow |
| Reconstruct ancestral WGD genome architecture | POInT | Order-aware ancestral reconstruction |
| Test post-WGD bias in fractionation (dosage-balance) | DupGen_finder + per-gene functional annotation | Compare retained vs lost genes |
| Distinguish recent vs ancient WGD | KsRates + Ks peak location after correction | Ks peak < 0.1 recent; 0.5-1.5 ancient |
| WGD in clade with rapid evolution (e.g. fish) | KsRates mandatory | Without rate correction, Ks peaks displaced |
| ML classification of WGD signal | SLEDGe | Newer ML-based alternative to manual fitting |
| Integrate WGD with DTL inference | Whale.jl OR ALE with WGD branch | See [[gene-tree-species-tree-reconciliation]] |
| Pan-clade WGD survey (e.g. all angiosperms) | wgd v2 batch + ksrates aggregated | Standard for sustained phylogenomic surveys |
| Polyploid genome with subgenomes | DupGen_finder for tandem/segmental + AnchorWave proali for subgenome-aware synteny | Subgenome assignment first |
| Recently diverged species pair, possible recent WGD | minimap2 -x asm5 + SyRI + Ks distribution | High-resolution recent-WGD detection |

## Per-Tool Failure Modes

### Saturation at Ks > 2

**Trigger:** Computing Ks for ancient WGD candidates from distantly related taxa.

**Mechanism:** Synonymous substitutions saturate after Ks ~2; each site has undergone multiple substitutions. Observed Ks underestimates true Ks; the relationship between Ks and time becomes non-monotonic above 1.5. WGD peaks at age 200+ Myr in vertebrates / 100+ Myr in plants are at Ks > 2 and unreliable (Vanneste 2013 GR 23:1304).

**Symptom:** wgd ksd output