Methods
Prototype: tiers rank annotated evidence, not proven effects. The rules and thresholds below are provisional and under review.
Evidence weights
A SNP's tier_score is the sum of the weights of its evidence, counted over all 21 conditions.
| Evidence | Weight | Fires when |
|---|---|---|
| active | 3 | An active state (Active TSS or Active enhancer) in at least one of the 21 conditions. |
| coding | 3 | A CDS change to the protein (missense, stop gained or lost, start lost) or of unknown effect. |
| synonymous | 3 | A synonymous CDS change (or stop retained): no amino-acid change (weight under review). |
| splice | 3 | In a splice region. |
| constrained | 2 | Inside a GERP constrained element (65 fish). |
| active_elsewhere | 1 | An active state only outside the chosen conditions (snpscore's --trait; never in the portal, which counts all 21). |
| weak | 1 | A weak state in at least one condition, and no active state in any. |
| utr | 1 | In a 5' or 3' UTR. |
| regbuild | 1 | Inside an Ensembl regulatory feature with no AQUA-FAANG peak. |
| motif | 1 | Inside a robust ATAC peak, one allele matches a JASPAR 2024 vertebrate motif at p < 1e-5 and the other doesn't, at least 4 bits (16-fold) apart. |
| molqtl | 0 | A molQTL lead SNP: shown, not scored. |
Coding consequences use Ensembl VEP terms on the canonical (else longest) Ensembl 116 transcript.
Tiers
| Tier 1 | tier_score 5 or more |
| Tier 2 | tier_score 3 to 4 |
| Tier 3 | tier_score 1 to 2 |
| Tier 4 | tier_score below 1: no evidence found (not evidence of no effect) |
Conditions
Tiers count chromatin activity in all 21 conditions: tissues (brain, gonad, liver, muscle; immature or mature, female or male) and the 5 embryo stages. The Table's Chromatin filter then keeps SNPs in a robust ATAC peak (or an active state) in the conditions you pick, without changing any tier. The portal names sample conditions only, never a phenotype.
Chromatin states
AQUA-FAANG's ChromHMM state labels, known only inside robust ATAC peaks. BodyMap (adult tissues) and DevMap (embryo stages) are separate 12-state models. The states have no written definitions yet: see the emission figure (ATAC, H3K27ac, H3K4me1, H3K4me3 and H3K27me3 signal per state).
| State | Model | Counts as |
|---|---|---|
| TSS Active_TSS | BodyMap and DevMap | active |
| Fl Flanking_upstream | BodyMap and DevMap | weak |
| Fl Flanking_downstream | BodyMap and DevMap | weak |
| Enh Active_enhancer | BodyMap and DevMap | active |
| Pri Primed_enhancer | BodyMap and DevMap | weak |
| Poi Poised_enhancer_1 | BodyMap and DevMap | weak |
| Poi Poised_enhancer_2 | BodyMap only | weak |
| Acc Accessible_chromatin | BodyMap and DevMap | weak |
| Biv Bivalent | DevMap only | weak |
| Pc Repressed_polycomb | BodyMap and DevMap | nothing |
| ZNF ZNF_genes_and_repeats | BodyMap and DevMap | nothing |
| Q Quiescent | BodyMap and DevMap | nothing |
| ? Unknown_1 | BodyMap and DevMap | nothing |
| no peak | No AQUA-FAANG peak in that condition. | nothing |
Glossary
- Robust ATAC peak
- An AQUA-FAANG open-chromatin peak reproduced across samples. States are only known inside these peaks.
- Blacklisted
- A region where peaks can't be called: no peak there means 'can't tell', not 'inactive'.
- LORe
- Region where the duplicated chromosomes still pair; genotyping probes may cross-hybridise.
- A/T or C/G SNP
- Reads the same on both strands; needs care on arrays.
- Flank variants
- Other public variants within 35 bp: they can disturb a genotyping probe.
- MAF
- Minor allele frequency in 504 public fish, wild and farmed, pooled; your strain can differ.
- molQTL lead SNP
- The SNP most associated with a gene's expression (eQTL) or a peak's accessibility (caQTL) in 12 AQUA-FAANG fish. Lead SNPs only, not fine-mapped; display only.
- cM
- Position on the Brekke et al. 2023 linkage map (AquaGen), interpolated between markers.
- Known locus
- A published SNP from the catalogue within 50,000 bp of yours.
- Ss4R duplicate
- The other copy of a gene from the salmonid genome duplication (Ss4R), from the AQUA-FAANG salmon-trout clades; also called its ohnolog.
- Coordinates
- Ssal_v3.1, chromosomes 1–29. Layers use BED intervals; a SNP at 1-based pos is inside one when start < pos ≤ end.
- TF motif change
- One allele of a SNP in a robust ATAC peak matches a transcription factor motif (JASPAR 2024, p < 1e-5) and the other doesn't, at least 4 bits apart; motifs are predicted binding, not proof of it.
- Synthetic
- Built from synthetic test data: not real results, never counted in tiers.
Fine-mapping of GWAS loci
Loci: SNPs within ±1,000,000 bp of each lead with p < 0.00001, strongest first, by distance only (no LD). Each locus is fine-mapped assuming one causal variant and no LD: Wakefield approximate Bayes factors with W = 0.2² on the standardised scale (betas divided by a trait SD of 1.0; without se, V = 1/(2N·af·(1−af))). The uniform prior gives every SNP the same weight; the functional prior π ∝ exp(0.5 × tier_score) over all 21 conditions. 95% credible sets in descending PIP.
Alleles are turned to the Ssal_v3.1 reference (alt = effect allele) using the genome, else the 504-fish panel sites; p values are read as text, so p = 1e-400 stays 400 on the −log10 scale.
All settings
The release's weights.yaml; every download records the checksum of the one it used.
Assay settings
| flank | 35 |
|---|---|
| gc | 0.3, 0.7 |
| max_repeat | 0.9 |
| max_homopolymer | 6 |
| max_flank_variants | 0 |
| warn_flags | lore, blacklist, strand_ambiguous |
| fail_ref_check | mismatch, not_snp |
Design settings
| n_target | 384 |
|---|---|
| backbone_fraction | 0.5 |
| min_spacing | 20000 |
| min_maf | 0.05 |
| max_per_gene | 2 |
| max_per_element | 1 |
| require_assay_pass | False |
| exclude_flags | blacklist |
| score | tier = 1.0; assay = 0.3; maf = 0.3 |
| chip_window | 10000 |
| chip_gap | 1000000 |
Finemap settings
| prior_sd | 0.2 |
|---|---|
| functional_lambda | 0.5 |
| coverage | 0.95 |
| significant | 5e-08 |
| suggestive | 1e-05 |
| window | 1000000 |
| near_lead | 2 |
| trait_sd | 1.0 |
Priors settings
| order | tier1_2, active, coding, constrained, regulatory |
|---|
Methods text for a paper
Show the text and references
SNPs were annotated with snpscore 0.1.0 (DuckDB 1.2.2) against release 2026-10-20 of the salmobase functional SNP annotation (assembly Ssal_v3.1, built 2026-10-20). Annotation layers: chromatin: AQUA-FAANG robust ATAC peaks with the ChromHMM state in each of 21 conditions (empty = no peak there); region: Genic region class from Ensembl 116 transcripts (CDS, splice region, UTR, non-coding exon, intron; empty = intergenic); regbuild: Ensembl 116 regulatory build (promoter, enhancer, open chromatin; promoters name their gene); gerp: GERP constrained elements from the Ensembl 116 65-fish alignment (fetch converts the bigBed to BED); blacklist: AQUA-FAANG blacklist; no peak here means "can't tell", not "no function"; lore: Late-rediploidized (LORe) regions, where the two Ss4R copies are near-identical; gene: Nearest Ensembl 116 gene (distance 0 = inside the gene); description is empty for many genes; gene_ortholog: Human and zebrafish ortholog gene names of the nearest gene (comma-separated when several); ss4r: Ss4R duplicate (ohnolog) of the nearest gene, its chromosome and rediploidisation class (AORe or LORe); flank_variants: Known variants within 35 bp either side (Axiom probe flank), from the public 504-fish panel (7,572,808 SNPs with MAF > 0.01); public_af: Public site at this exact position in the 504-fish panel: ALT allele frequency (af), minor allele frequency (maf), ALT allele count (ac), alleles (an) and fish (ns) with data; aquafaang12_af and aquafaang12_dr2 are the ALT frequency and imputation quality in the 12 AQUA-FAANG fish only (NULL where the site is not in their VCF); synchrosmolt_af: Site at this exact position among the 108 whole-genome sequenced parents of the Synchrosmolt family experiment: ALT allele frequency (af), minor allele frequency (maf), ALT allele count (ac), alleles (an) and fish (ns) with a call (411,908 SNPs, phased with Beagle); motif: TF motif changes at this site (inside a robust ATAC peak): changed (one allele matches a JASPAR 2024 vertebrate motif at p < 1e-5 and the other doesn't, at least 4 bits apart), n (motifs changed), tfs ('CTCF (G), FOXO3 (A)': the TF and the allele that matches, largest change first) and delta (the largest score difference, bits); molqtl: Lead SNP of a caQTL (ATAC peak; Brain, Gonad, Liver, Muscle) or eQTL (gene; Brain, Gill, Gonad, Liver, Muscle) in the nf-RASQUAL analysis of the 12 AQUA-FAANG fish, with eigenMT-adjusted p (BF) below max_bf: n features, tissues, kinds (ca, e), best_bf and features ('<tissue> <kind> <gene or peak> (BF, pi)'; genes are Ensembl 106 IDs, peaks chrom:start-end in BED coordinates). Lead SNPs only (one per feature, 12 fish: low power, winner's curse); pi is RASQUAL's effect size, its allele direction is not confirmed. Gill ATAC is left out (the assay was discarded); cm: Genetic map position in cM (male, female), interpolated between the flanking markers of the Brekke et al. 2023 linkage map (AquaGen breeding population, 34,842 markers on Ssal_v3.1, Haldane); cm is false and the values NULL beyond a chromosome's first and last marker. The male map misses crossovers on chromosomes 2, 8 and 17 (19.5, 2.0 and 3.8 cM in total), and unmarked chromosome ends (up to 12 Mb) have no value; expression: Expression of the nearest gene, mean counts per million per tissue (nf-rasqual RNA-seq, 12 fish per tissue).
Each SNP was given a functional tier (1 best to 4) counting all AQUA-FAANG conditions. tier_score adds the weights of the evidence it has: an active chromatin state (Active_TSS, Active_enhancer) in any condition (3), a weak state (1), a CDS change to the protein or of unknown effect (3), a synonymous CDS change (3), splice region (3), UTR (1), a GERP constrained element (2) and an Ensembl regulatory feature outside AQUA-FAANG peaks (1), and a transcription factor motif changed inside a robust ATAC peak (1: one allele matches a JASPAR 2024 vertebrate motif at p < 1e-5 and the other does not, at least 4 bits apart); tiers 1-3 need a score of at least 5, 3, 1. These rules are PROVISIONAL.
Coding consequences (Sequence Ontology terms as Ensembl VEP writes them, with HGVS protein changes) were computed from the genome sequence on one transcript per SNP (Ensembl release 116 gene annotation, Ssal_v3.1 (ensembl.org); the canonical transcript, else the longest CDS), after turning swapped or strand-flipped alleles to the reference; splice donor and acceptor sites are the two intronic bases at internal exon ends.
Assay designability was checked on 35 bp flanks either side of each SNP from the soft-masked Ssal_v3.1 genome, with neighbouring public variants written as IUPAC codes. SNPs failed on ref mismatch or not a SNP or any N in the flanks, and were flagged (warn) for GC outside 0.3-0.7, a soft-masked share above 0.9, a single-base run longer than 6, more than 0 public variants on both sides of the SNP (one clean flank is enough), an A/T or C/G SNP whose Illumina TOP/BOT strand cannot be determined from the flanks, or lore, blacklist, strand_ambiguous. These thresholds are PROVISIONAL, to confirm with the genotyping provider.
References
- AQUA-FAANG consortium, Atlantic salmon blacklist (salmobase.org) (DOI to add)
- AQUA-FAANG consortium, Atlantic salmon unified ATAC peaks and ChromHMM states (salmobase.org) (DOI to add)
- AQUA-FAANG tables, Ssal_late_rediploidized_regions.tsv (salmobase.org) (DOI to add)
- Baudement M-O, Perojil Morata D, Gillard GB, ..., Kent MP, Lien S, Macqueen DJ. Salmonids reveal principles of regulatory evolution following autotetraploidization. bioRxiv preprint 2025, v2 of 2025-12-23, not peer reviewed, doi:10.64898/2025.12.18.695160
- Davydov EV et al. (2010) Identifying a high fraction of the human genome to be under selective constraint using GERP++. PLoS Comput Biol 6: e1001025. doi:10.1371/journal.pcbi.1001025
- JASPAR 2024 CORE vertebrates, non-redundant (879 motifs) (DOI to add)
- Nguyen DT, Sandve SR, Lien S, Grønvold L. BMC Genomics 2025;26:1070, doi:10.1186/s12864-025-12302-5
- Rauluseviciute I et al. 2024, Nucleic Acids Research 52:D174, doi:10.1093/nar/gkad1059
- Saitou M et al. 2025, bioRxiv doi:10.1101/2025.10.16.682803
- Synchrosmolt parent SNP genotypes, ENA PRJEB114348 (NMBU, 2026 (DOI to add)
- nf-rasqual 504-fish reference panel (salmobase.org) (DOI to add)
- reads PRJEB47441) (DOI to add)
- sample collection: Bertolotti et al. 2020 (ENA PRJEB38061) (DOI to add)
- Known loci (IPN QTL (near nae1), LOC106600468 (DNA ligase 1-like), LOC106611243 (F-box only protein 34-like), MHC II region (Ssa12), MHC class I region (Ssa27), Ssa07 (array GWAS), Ssa07 PD QTL (Ig light chain region), Ssa12 QTL, Ssa16 (WGS GWAS), akap11, bco1l, cdh1-1 (LOC106587268), cntn4, gig1-like cluster (Ssa03 PD QTL), gig1-like copy 3, hubsmolt (chr14 hub locus), itprid2, magi2, ncor1 region (landlocked sweep), ndufs4, pi4kb/bnip2 region (Ssa27), picalm, rora, scaf11/slc38a2, sdY, six6, vgll3): doi:10.1016/j.ygeno.2021.09.012; doi:10.1038/nature16062; doi:10.1038/s41437-022-00570-w; doi:10.1038/s41598-019-56438-3; doi:10.1038/s41598-020-67405-8; doi:10.1093/jhered/esz042; doi:10.1101/2025.10.16.682803; doi:10.1186/1471-2164-13-244; doi:10.1186/s12711-020-0529-8; doi:10.1186/s12711-025-01026-5; doi:10.1186/s12863-024-01263-5; doi:10.1186/s12864-019-5525-4; doi:10.1186/s12864-019-6104-4; doi:10.1186/s12864-020-06788-4; doi:10.1186/s12864-025-11735-2; doi:10.1371/journal.pgen.1005628; doi:10.1534/g3.115.020115; doi:10.1534/genetics.115.175406; doi:10.3389/fgene.2020.544207; doi:10.3389/fgene.2024.1381333; doi:10.3390/genes11060608
- AQUA-FAANG consortium, Atlantic salmon unified ATAC peaks and ChromHMM states (salmobase.org); Baudement M-O, Perojil Morata D, Gillard GB, ..., Kent MP, Lien S, Macqueen DJ. Salmonids reveal principles of regulatory evolution following autotetraploidization. bioRxiv preprint 2025, v2 of 2025-12-23, not peer reviewed, doi:10.64898/2025.12.18.695160 (chromatin)
- Ensembl release 116 gene annotation, Ssal_v3.1 (ensembl.org) (region, gene, transcripts)
- Ensembl release 116 regulatory build (ensembl.org) (regbuild)
- Ensembl Compara release 116, 65 fish GERP constrained elements (ensembl.org) (gerp)
- AQUA-FAANG consortium, Atlantic salmon blacklist (salmobase.org); Baudement M-O, Perojil Morata D, Gillard GB, ..., Kent MP, Lien S, Macqueen DJ. Salmonids reveal principles of regulatory evolution following autotetraploidization. bioRxiv preprint 2025, v2 of 2025-12-23, not peer reviewed, doi:10.64898/2025.12.18.695160 (blacklist)
- AQUA-FAANG tables, Ssal_late_rediploidized_regions.tsv (salmobase.org); Baudement M-O, Perojil Morata D, Gillard GB, ..., Kent MP, Lien S, Macqueen DJ. Salmonids reveal principles of regulatory evolution following autotetraploidization. bioRxiv preprint 2025, v2 of 2025-12-23, not peer reviewed, doi:10.64898/2025.12.18.695160 (lore)
- Ensembl Compara release 116 orthologues, via BioMart (ensembl.org) (gene_ortholog)
- AQUA-FAANG salmon-trout orthologs (Ss4R quartets from Ensembl 106 gene trees), salmobase.org (ss4r)
- nf-rasqual 504-fish reference panel, sites only (salmobase.org) (flank_variants)
- nf-rasqual 504-fish reference panel (salmobase.org); Nguyen DT, Sandve SR, Lien S, Grønvold L. BMC Genomics 2025;26:1070, doi:10.1186/s12864-025-12302-5; sample collection: Bertolotti et al. 2020 (ENA PRJEB38061) (public_af)
- Synchrosmolt parent SNP genotypes, ENA PRJEB114348 (NMBU, 2026; reads PRJEB47441); Saitou M et al. 2025, bioRxiv doi:10.1101/2025.10.16.682803 (synchrosmolt_af)
- JASPAR 2024 CORE vertebrates, non-redundant (879 motifs); Rauluseviciute I et al. 2024, Nucleic Acids Research 52:D174, doi:10.1093/nar/gkad1059 (motif)
- Nguyen DT, Sandve SR, Lien S, Grønvold L. BMC Genomics 2025;26:1070, doi:10.1186/s12864-025-12302-5 (tables from github.com/datngu/nf-RASQUAL_paper) (molqtl)
- Brekke et al. 2023, Sci Rep, doi:10.1038/s41598-023-47208-3 (map from bioRxiv 2023.06.07.543993 v2, supplementary media-1.txt) (cm)
- nf-rasqual RNA-seq gene counts (salmobase.org) (expression)
- NCBI RefSeq annotation GCF_905237065.1-RS_2025_05, Ssal_v3.1 (ncbi.nlm.nih.gov) (transcripts_ncbi)