skills.rest
mims-harvard
mims-harvardOFFICIAL
Explore all skills by this author →
Website𝕏@marinkazitnikGitHub

tooluniverse-peptide-target-deorphanization

Identify the real protein targets of a peptide from its sequence and phenotype.

1.7k|254|Updated Mar 3, 2025
Search Tags:#bioinformatics#drug discovery#peptide#co-folding#gpcr#protein target identification#target deorphanization
One-click install
npx skills add https://github.com/mims-harvard/ToolUniverse --skill tooluniverse-peptide-target-deorphanization
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: tooluniverse-peptide-target-deorphanization
Source: https://github.com/mims-harvard/ToolUniverse/tree/main/plugins/tooluniverse/skills/tooluniverse-peptide-target-deorphanization
Command: npx skills add https://github.com/mims-harvard/ToolUniverse --skill tooluniverse-peptide-target-deorphanization

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires tooluniverse, and includes scripts (resource) and references (resource) components.

What problem does it solve?

A peptide can show a clear phenotype yet fail to bind its hypothesized target, leaving researchers guessing which protein it actually engages. This Skill deorphanizes peptides by combining sequence, motif, homology, target-family, phenotype, and cross-species evidence to rank the most likely real targets instead of relying on name-level assumptions.

Core Features & Use Cases

  • Multi-route candidate generation: Unions BLAST homology, PROSITE/ELM motif signatures, HGNC/InterPro/GPCRdb target-family enumeration, and OpenTargets phenotype anchoring into one ranked shortlist.
  • Target-class router: Classifies the peptide (GPCR ligand, ion-channel toxin, protease target, cytokine receptor, integrin ligand, and more) so enumeration adapts to any target class, not just GPCRs.
  • Cross-species reconciliation: Aligns human, assay-species, and source-species ortholog sequences to explain "binds in species A but not B" results, plus DPP4/protease liability flags.
  • Optional structural confirmation: Co-folds the peptide against shortlisted receptors via NVIDIA NIM (Boltz2, AlphaFold2-Multimer, OpenFold3) ranked by interface ipTM, with a keyless dry-run mode.
  • Use Case: Given exendin-4 with a negative GLP1R binding result and a type 2 diabetes phenotype, the pipeline recovers the class-B GPCR panel and promotes GIPR to Tier 1 as the leading real-target hypothesis, with zero API keys.

Quick Start

Run the deorphanization pipeline on my peptide sequence HGEGTFTSDLSKQMEEEAVRLFIEWLKNGGPSSGAPPPS with hypothesized target GLP1R and phenotype type 2 diabetes mellitus to find its real target.

Frequently Asked Questions about tooluniverse-peptide-target-deorphanization

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I find the real target of a peptide that does not bind its hypothesized receptor?▼

Run deorphanize_peptide.py with the peptide sequence, the hypothesized target gene, and the disease phenotype. It unions homology, motif, target-family, and OpenTargets phenotype evidence into a ranked candidate shortlist, promoting paralogs and phenotype-shared receptors when the hypothesized target tests negative.

Can peptide target deorphanization work for non-GPCR targets like ion channels or proteases?▼

Yes. A target-class router classifies the peptide as a GPCR ligand, ion-channel toxin, protease substrate, cytokine, integrin ligand, or other class, then adapts family enumeration using HGNC gene families and InterPro, which work for any target class. GPCRdb is only a GPCR-specific cross-check.

Does the deorphanization pipeline require an API key?▼

No key is needed for candidate generation and ranking, which use keyless tools like BLAST, PROSITE, HGNC, OpenTargets, and UniProt. Only the optional Phase 5 co-folding confirmation requires NVIDIA_API_KEY, and cofold_screen.py runs a dry run without it.

Why does my peptide bind in one species but not another?▼

Species-specific binding usually reflects divergence in the receptor's ligand-binding interface, not a different target. The pipeline resolves human, assay-species, and source-species ortholog sequences via UniProt and aligns them, reporting percent identity to flag the diverged ortholog.

What are the limitations of sequence-based peptide target prediction?▼

BLAST, PROSITE, and ProtParam assume canonical linear peptides, so non-ribosomal or cyclic peptides may be mischaracterized and should be checked with Norine and co-folded with the cyclic flag. Single-route candidates remain Tier 3 hypotheses until corroborated by phenotype or structural evidence.

Related Skills

Explore More Education & Research Skills→
felixboehmfelixboehm

adaptyv

Submit protein sequences for automated experimental assays and computational optimization.

Community
Advanced
hxk622hxk622
8

adaptyv

Submit protein sequences for automated binding assays and validation.

Community
Advanced
jaechang-hitsjaechang-hits
298

gtopdb-database

Query the GtoPdb REST API for receptor-ligand interaction data.

Community
Intermediate
inflexa-aiinflexa-ai
29

proteomics

Perform differential expression analysis on quantitative proteomics data using limma and MSstats.

Official
Advanced