Lake Trout EpiGenomics Project
From Methylation & Structural Variation to Candidate Genes — Salvelinus namaycush Ecotypes
Two Ecotypes, One Lake System
Lake trout (Salvelinus namaycush) in the Great Lakes occur as divergent ecotypes that share water but not lifestyle:
- Lean: shallow-water dwelling, elongate body, low lipid content
- Siscowet: deep-water specialist, robust body, high lipid storage
This project asks a focused question: which genes carry the epigenetic and structural differences between the ecotypes, and what phenotypes might they shape? Earlier stages produced the raw differences — differentially methylated regions (DMRs) and presence–absence variants (PAVs). The work featured here adds the missing interpretive layer: a genome-wide functional annotation that turns coordinates into gene names, products, and Gene Ontology terms, then ranks candidates and reasons — carefully — about phenotype.
Read this as hypothesis-generating. Every link below is an association on a single lean-background reference genome — no functional validation, and no single CpG survives genome-wide multiple-testing correction. The value is a ranked, annotated shortlist, not a causal claim.
How the Evidence Stacks: Three Integrated Layers
Differential methylation
- 540,040 CpG sites tested
- 302 DMRs (20 hyper- / 282 hypo-methylated in siscowet)
- 149 DMRs within 5 kb of a gene; 88 in promoters
- 0 single CpGs survive q < 0.1 — lead with the DMR level
Presence–absence variation (PAV)
- 3,465 stringent siscowet-specific deletions (>100 bp, all-4-vs-none)
- 1,543 within 5 kb of a gene; 54 overlap an exon (candidate copy/LOF changes)
- Reference-bias aware: a lean-background genome inflates siscowet deletions
Functional annotation (new)
- 46,359 genes annotated from NCBI RefSeq
- 46,231 with a product description; 34,367 with ≥1 Gene Ontology term
- The join key that turns variants into interpretable candidates
See analyses/18-annotation/README.md for the annotation methods and provenance.
Convergent & Top-Ranked Candidate Genes
Genes were ranked by convergence (methylation and deletion), promoter/exon placement, expression support, and methylation↔︎expression concordance. The four convergent loci — carrying both a DMR and a high-confidence siscowet deletion — are the strongest candidates.
| Gene | Product | Methylation | Deletion | Note |
|---|---|---|---|---|
znf883-like (LOC120032414) |
Zinc finger protein 883-like | exon · hyper | exonic | top convergent |
XlCGF57.1-like (LOC120040411) |
Gastrula zinc finger protein XlCGF57.1-like | intron · hypo | nearby | convergent |
septin-9-like (LOC120043843) |
Septin-9-like | intron · hyper | nearby | convergent |
| LOC120039781 | Uncharacterized locus | intron · hypo | nearby | convergent |
angptl5 |
Angiopoietin-related protein 5-like | — | exonic | lipid axis |
mogat2 |
2-acylglycerol O-acyltransferase 2-A-like | — | exonic | lipid axis |
ephx1-like |
Epoxide hydrolase 1-like | promoter | — | lipid / xenobiotic |
Source: integrated_candidate_genes.tsv.
Gene Ontology enrichment (deletion set)
| GO term | Fold | FDR | Read as |
|---|---|---|---|
| Calcium ion transmembrane transport | 3.7 | 5.6×10⁻⁴ | most defensible signal |
| Neuron projection development | 2.4 | 2.6×10⁻³ | sensory / neural |
| Calcium channel complex | 4.6 | 3.0×10⁻³ | length-bias caveat |
| Calcium ion transport | 3.0 | 3.0×10⁻³ | ion homeostasis |
| Lipid / phospholipid binding | 1.5 | ns (0.3) | suggestive only |
Hypergeometric over-representation vs. all GO-annotated genes (BH-FDR). The DMR set’s enrichment is dominated by a single histone cluster and adjacent znf883 paralogs — a tandem-cluster artifact, not broad convergence. Full tables: PAV, DMR, union.
Hypothesized Links to Ecotype Biology
Each axis pairs annotated candidate genes with a known or measured difference between the ecotypes. These are hypotheses anchored to morphometric data, not validated mechanisms.
- 🫧 Lipid & energy storage — exonic siscowet deletions in
angptl5,mogat2, and a promoter DMR at epoxide hydrolase 1 touch lipid handling, consistent with the defining high-lipid siscowet phenotype and deep-water buoyancy. - 🐠 Body shape, growth & muscle — methylation-led candidates including
rbm24b(muscle/cardiac splicing) and growth-associated GO terms align with the elongate-lean vs. robust-siscowet contrast captured in the 17-landmark morphometric data. - ⚡ Calcium / sensory-neural — calcium-transport and neuron-projection GO enrichment in the deletion set hints at sensory or excitability differences relevant to a deep, dark, high-pressure habitat (length-bias-aware).
- 🛡️ Immune & adhesion — exonic deletions hit immune/adhesion genes (alpha-2-macroglobulin, CEACAM, DMBT1); some signal is real divergence, some reflects rapidly evolving, reference-divergent families.
Full write-up: code/18-diff-annotation-phenotype.html.
Interactive Genome Browsers
Explore methylation, PAV, gene, and ecotype-synteny tracks directly across the SaNama_1.0 assembly. Genes now carry functional annotation (symbol, product, GO terms, and which ecotypes retain them), and lean/siscowet synteny blocks are projected onto the reference so you can see, at any locus, which ecotype contig maps there and whether it is inverted.
🔬 IGV.js — quick exploration
Functionally-annotated genes · lean & siscowet synteny blocks · PAV insertions & deletions · CpG methylation (8 samples) · DMRs
🧬 JBrowse 2 — advanced analysis
Functionally-annotated genes (GFF3) · lean & siscowet synteny blocks · PAV structural variants · CpG methylation · differential methylation · lean ↔︎ siscowet Linear Synteny View
For the side-by-side ecotype comparison, open JBrowse’s Linear synteny view and pick the lean_purged and siscowet_purged assemblies with the Lean ↔︎ Siscowet synteny track.
Interpretation Guardrails
This analysis is deliberately conservative. The constraints below shape every claim above and are baked into the candidate rankings.
The reference is a lean-background genome. SaNama_1.0 was built from a doubled-haploid Seneca Lake (lean-morphotype) fish. Siscowet diverges more from it, so siscowet reads map less completely — inflating apparent siscowet-specific deletions and reducing methylation power in the most divergent regions. Siscowet and lean variant counts are not magnitude-comparable.
No single CpG survives genome-wide correction (0 DMCs at q < 0.1). Interpretation leads with DMR-level and stringent-PAV sets; single-CpG and lenient-PAV hits are hypothesis-generating only.
Expression support is weak by design. The liver RNA-seq is from a separate parasite study with different individuals — orthogonal support, never confirmation.
Enrichment confounders. The PAV GO signal carries a gene-length bias (long calcium/ion-channel genes accumulate deletions by chance); the DMR GO signal is a tandem-cluster artifact. Associations, not causation — no functional validation was performed.
Data & Methods
Reference Genome
- Assembly: NCBI GCF_016432855.1 (SaNama_1.0)
- Source: NCBI BioProject PRJNA674328
- Annotation: NCBI RefSeq GFF + GO (GAF)
Samples
| Ecotype | Sample Size | Description |
|---|---|---|
| Lean | n=4 | Shallow-water ecotype (PacBio HiFi) |
| Siscowet | n=4 | Deep-water ecotype (PacBio HiFi) |
Analysis Pipeline
- PacBio HiFi sequencing with 5mC modification calling
- CpG methylation profiling & DMR identification
- Coverage/CIGAR-based PAV detection (lenient + stringent tiers)
- RefSeq functional annotation backbone (gene → product → GO)
- Strand-aware DMR/PAV-to-gene assignment (promoter ±2 kb, flank ±5 kb)
- Hypergeometric GO over-representation with BH-FDR
Citation & Resources
- GitHub Repository: RobertsLab/project-lake-trout
- Annotation plan:
code/18-diff-annotation-phenotype-plan.md