Interactions & Expression

Protein interactions, dimer interfaces, and cell-type expression.

These panels place each isoform in its cellular context by integrating gene-level protein interaction networks (STRING), splice-driven homodimer and heterodimer interface disruptions, and single-cell expression profiles (Human Protein Atlas).

STRING Interaction Evidence

Highest-confidence gene-level STRING partners are shown (keyed by Ensembl protein ID). The combined score is decomposed into seven underlying evidence channels:

Experiments

Direct biochemical / biophysical assays.

Database

Curated pathway & complex memberships.

Co-expression

Correlated expression across conditions.

Text-mining

Co-mention in the literature.

Phylogenetic profile

Co-occurrence across genomes.

Gene fusion

Fused orthologs in other species.

Neighborhood

Conserved genomic proximity (prokaryotes).

Note: STRING evidence is gene-level and does not distinguish isoform-specific interactions. High combined scores may be driven by co-expression or text-mining rather than physical binding.

Homodimer Interface Analysis

For genes with a predicted homodimer interface, canonical interface residues are mapped onto the novel isoform. The disruption_score quantifies the impact of splicing by scoring absent and conformationally perturbed residues (SASA change).

Note: The boolean flag interface_disrupted_by_deletionis TRUE when ≥ 50% of the canonical interface residues are physically absent from the isoform. Use this to separate total interface deletion from conformational disruption.

Heterodimer Interface Analysis

The same interface logic is applied to heterodimers — complexes with a distinct partner protein. Interface residues are taken from the AlphaFold Database model of the pair (afdb_accession) and mapped onto the isoform, with count_absent and count_sasa_changed feeding the same disruption_score. This captures splice events that break a specific binding partnership rather than self-association.

Each row names the partner (partner_name / partner_uniprot_id), so an isoform can be read as retaining some interfaces while losing others.

Cell-type Expression (HPA)

The expression profile covers 154 cell types from the Human Protein Atlas (HPA v25). Each entry includes the nCPM expression value and the specificity category (e.g., cell_type_enriched).

Note: HPA data is aggregated at the gene level. Isoform-specific expression abundance cannot be directly inferred from this profile.

Interactive Tissue Visualization

Every isoform page features a dynamic human body visualization in the top bar. This interactive map highlights tissues based on two distinct data sources:

  • Long-read Cohorts (Colored): Tissues directly confirmed by SPLISOFORMS long-read sequencing cohorts (e.g., Breast Cancer, ccRCC Kidney) are highlighted with bold primary colors, solid callout lines, and pulsing animated indicators.
  • HPA Baseline (Grey): The standard 11 baseline organs tracked by the Human Protein Atlas are mapped statically in the background as a grey baseline to provide spatial context for expression.