Skip to main content

​
Overview

TRIFID integrates data from multiple external databases to generate comprehensive predictive features for splice isoform functionality assessment. This page describes each data source and how to access them.

​
Primary Annotation Sources

​
GENCODE

Description: The reference human gene annotation database providing comprehensive gene and transcript annotations. Usage in TRIFID: Source of protein-coding transcript annotations, exon structures, and genomic coordinates. Access: File Types Used:
  • GTF annotations: gencode.vXX.annotation.gtf.gz
  • GFF3 annotations: gencode.vXX.annotation.gff3.gz
TRIFID was initially trained on GENCODE Release 27 (GRCh38.p10) but now supports multiple releases.

​
Ensembl

Description: Genome browser and annotation database for vertebrate and model organism genomes. Usage in TRIFID: Primary annotation source for non-human species including rat, zebrafish, chicken, and other vertebrates. Access:
  • Website: ensembl.org
  • FTP Server: Available through Ensembl FTP
Species Supported: Rat, zebrafish, chicken, chimpanzee, pig, cow, macaque, fruitfly, and worm.

​
RefSeq (NCBI)

Description: NCBI Reference Sequence database providing curated genomic, transcript, and protein sequences. Usage in TRIFID: Alternative annotation source for human genome (GRCh37 and GRCh38). Access:

​
Protein Structural and Functional Data

​
APPRIS

Description: Database annotating principal and alternative splice isoforms using protein structural and functional information. Usage in TRIFID: Provides multiple features including:
  • Principal isoform annotations
  • Protein structural integrity scores
  • Functionally important residue annotations
  • Cross-species conservation evidence
  • ALT-Corsair evolutionary age scores
Access: File Types Used:
  • appris_data.principal.txt: Principal isoform labels
  • appris_data.appris.txt: APPRIS method scores
  • appris_data.transl.fa.gz: Protein sequences
  • appris_method.spade.gtf.gz: Pfam domain annotations
Download Example (GENCODE 27): 
mkdir -p data/external/appris/GRCh38/g27
curl http://apprisws.bioinfo.cnio.es/pub/current_release/datafiles/homo_sapiens/e90v35/appris_data.principal.txt \
  -o data/external/appris/GRCh38/g27/appris_data.principal.txt

​
Pfam

Description: Database of protein families and domains. Usage in TRIFID: Used to quantify the impact of alternative splicing on protein domains. TRIFID’s Pfam effects module calculates whether domains are damaged, lost, or intact. Access: Pre-computed Scores:

​
Evolutionary Conservation Data

​
PhyloCSF

Description: Tool for measuring evolutionary conservation of coding sequences using codon substitution frequencies. Usage in TRIFID: Complementary measure of evolutionary conservation across species. Access:

​
ALT-Corsair

Description: APPRIS module that quantifies the evolutionary age of orthologs by identifying the last common ancestor of the most distant ortholog. Usage in TRIFID: Reports evolutionary age scores representing conservation across species. Access:

​
Expression Data

​
E-MTAB-2836 (RNA-seq)

Description: Large-scale RNA-seq expression dataset from 122 human individuals across 32 tissues. Usage in TRIFID: Source data for QSplice module to quantify splice junction coverage. Access: Pre-computed Scores:
Processing RNA-seq data requires the APPRIS RNA-seq pipeline.

​
Training Data

​
Proteomics Evidence (Kim et al., 2014)

Description: Large-scale proteomics study providing experimental evidence for protein isoforms. Usage in TRIFID: Gold standard training set for the machine learning model. Citation: Kim et al., 2014, Nature Pre-computed Training Set:

​
Pre-computed TRIFID Module Outputs

For convenience, pre-computed outputs from TRIFID modules are available:
ModuleDescriptionData Link
QSpliceSplice junction coverage scoresDownload
Pfam EffectsDomain impact scoresDownload
Fragment LabellingRedundancy labelsDownload
PhyloCSFConservation scoresDownload
ALT-CorsairEvolutionary age scoresDownload

​
Centralized Data Repository

All source data files needed to reproduce TRIFID analysis are available in the centralized sharepoint: Google Drive - TRIFID Source Data

​
Feature Documentation

Detailed descriptions of all 47 predictive features are available: Feature Documentation PDF

Build docs developers (and LLMs) love

Get started for freeTalk to us