Module Species Compatibility

Full

The underlying resource natively covers the species. Results are statistically supported.

Ortholog

Computed on one-to-one human orthologs. Usable, with reliability decreasing as phylogenetic distance grows.

Limited

Annotation coverage is partial for the species or its relatives. Interpret with caution.

Hypothesis

No direct annotation support. Outputs are exploratory, hypothesis-generating leads only.

Why applicability differs across species

Origin of the tools: several integrated tools — CellPhoneDB, NicheNet, SCENIC, DisGeNET, OMIM disease associations and drug–target annotations — were developed primarily from human or mammalian annotations.
LACA's cross-species route: for non-human species, genes are converted to one-to-one human orthologs (OrthoFinder, with BioMart and DIAMOND reciprocal-best-hit fallbacks) before these modules are run. This is the practice recommended by the tool developers themselves — e.g. the CellPhoneDB documentation instructs users of non-human data to map gene identifiers to human orthologs, NicheNet's prior model was built from human and mouse data, and SCENIC has been applied to non-mammalian species such as Drosophila.
Transparent coverage: per-species ortholog coverage is quantified and the complete mapping — flagged by mapping type (one-to-one, one-to-many, missing) — is downloadable from the Download Center, so users can see exactly how many genes of a species enter each analysis. The per-species summary (assignment counts and one-to-one fractions, with the mapping method used) is directly available as ortholog_mapping_summary.csv, and the full gene-level mapping as ortholog_mapping_full.csv.gz.
Guardrail: an annotation-dependent module is not run for a species when its underlying resources do not cover that species; where outputs are produced through ortholog projection, module pages label them accordingly.

Outputs of annotation-dependent modules in non-human species are hypotheses rather than statistically supported conclusions. In particular, disease phenotypes are not equivalent across species: a disease association reported for a non-human species indicates a shared molecular etiology candidate, not the presence of the disease.

Module	Annotation basis	Human	Non-human mammals	Non-mammalian vertebrates	Invertebrates	Cross-species strategy	Interpretation guidance
Tier 1 — Data-driven modulesNo external annotation required; statistically supported for all 134 species.
Co-expression analysisIn-house correlation pipeline	None — computed from the expression data of the queried species.	Full	Full	Full	Full	Not required; computed within each species on native expression values.	Statistically supported for all 134 species.
Trajectory analysisPAGA; Monocle 2	None — data-driven topology and pseudotime inference.	Full	Full	Full	Full	Not required.	Statistically supported for all 134 species; the biological direction of pseudotime requires external markers.
DEG analysisIn-house statistical testing, FDR-corrected	None — data-driven differential testing.	Full	Full	Full	Full	Not required.	Statistically supported for all 134 species.
Tier 2 — Annotation-dependent modules, extensible by orthologyFully supported for human; extended to other species through one-to-one human ortholog mapping, with reliability decreasing with phylogenetic distance.
Cell communication analysisCellPhoneDB; NicheNet	Human-curated ligand–receptor pairs (CellPhoneDB); ligand–target prior model built from human and mouse data (NicheNet).	Full	Ortholog	Ortholog	Hypothesis	Non-human genes are converted to one-to-one human orthologs before analysis, following the tool developers’ recommended practice; an interaction is reported only when both partners are mapped.	Statistically supported for human; for non-human species, interactions are candidates requiring experimental validation.
GRN (gene regulatory network) analysisSCENIC	TF–motif annotations (cisTarget motif databases) available primarily for human, mouse and fly.	Full	Ortholog	Limited	Limited	Regulons are inferred on ortholog-mapped genes using the closest available motif database.	Regulons inferred for non-model species are hypothesis-generating and should be verified against species-specific regulatory evidence.
GO/KEGG enrichment & GSEAGO; KEGG; GSEA	GO annotations (most complete for human, mammals and model organisms); KEGG KO ortholog groups (species-aware); MSigDB gene sets (human-centric).	Full	Ortholog	Ortholog	Limited	Non-human genes are analyzed through KO assignments or one-to-one human orthologs.	Enrichment results for non-model species are affected by annotation coverage bias; interpret negative results with caution.
Motif enrichment analysisIn-house motif scanning (JASPAR-style PWMs)	TF binding motifs derived mainly from human and mammalian experiments; PWMs broadly conserved across vertebrates.	Full	Ortholog	Ortholog	Limited	Motifs are matched within the ortholog-mapped TF/target space.	Motif enrichment in distantly related species is hypothesis-generating.
Metabolism analysisIn-house pathway-activity scoring (KEGG/Reactome gene sets)	Human and mammalian enzyme and pathway annotations.	Full	Ortholog	Ortholog	Limited	One-to-one human ortholog mapping.	Pathway-activity estimates in distantly related species are hypothesis-generating.
Tier 3 — Human-annotation-bound modulesNative to human; for any non-human species these modules operate through human orthologs only and must be regarded as hypothesis-generating.
ToppGene enrichmentToppGene	Human gene functional annotations and curated human gene sets.	Full	Ortholog	Ortholog	Hypothesis	One-to-one human ortholog mapping required.	Results for non-human species are hypothesis-generating.
Disease association analysisDisGeNET; OMIM	Human disease–gene associations only.	Full	Ortholog	Ortholog	Hypothesis	Non-human genes are mapped to human orthologs; disease phenotypes are not equivalent across species.	Associations in non-human species indicate shared molecular etiology candidates rather than the presence of the disease; hypothesis-generating.
Drug target analysisIn-house integration of compound–target annotations	Human-centric pharmacological compound–target annotations.	Full	Ortholog	Ortholog	Hypothesis	One-to-one human ortholog mapping required.	Drug–target calls in non-human species are hypothesis-generating and mainly inform translational prioritization toward human.

Why applicability differs across species

Related resources