protein-sequence-similarity-search

Search protein sequence databases for homologues and infer functions.

2.7k|283|Updated May 13, 2026
One-click install
npx skills add https://github.com/google-deepmind/science-skills --skill protein-sequence-similarity-search
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: protein-sequence-similarity-search
Source: https://github.com/google-deepmind/science-skills/tree/main/skills/protein_sequence_similarity_search
Command: npx skills add https://github.com/google-deepmind/science-skills --skill protein-sequence-similarity-search

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires scienceskillscommon, python-dotenv, and includes scripts (resource) components.

What problem does it solve?

It identifies homologous protein sequences and helps infer likely protein functions from a user-provided amino-acid sequence, FASTA file, or related query.

Core Features & Use Cases

  • Fast homologue search (MMseqs2 via ColabFold API): Runs a quick search by default and produces a ranked Markdown table of top hits.
  • Comprehensive homologue search (EBI BLAST fallback): Falls back to slower BLAST when MMseqs2 fails (e.g., rate limits) or when the user explicitly requests BLAST or specific UniProt/structure databases.
  • Functional interpretation and saved outputs: Interprets alignment metrics (Q-Cov, E-value, sequence identity) and writes results to local .md plus optional .json files for follow-on analysis.

Quick Start

Tell your agent: "Run a fast homologous protein search for this FASTA file and summarize the top hits and likely functions."

Frequently Asked Questions about protein-sequence-similarity-search

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

FAQPage Schema
How do I find homologous proteins for a raw amino-acid sequence?

You can find homologous proteins by providing a raw amino-acid sequence or FASTA file as input. The tool searches sequence similarity databases using MMseqs2 to identify matching protein homologues and infer protein function.

What is the difference between MMseqs2 and BLAST for protein function inference?

MMseqs2 provides fast homologue searching via the ColabFold API, while BLAST offers a slower, comprehensive fallback through EBI. BLAST is used when MMseqs2 hits rate limits or when specific UniProt or structure databases are explicitly requested.

How do I perform a homology search when MMseqs2 is rate-limited?

When MMseqs2 is rate-limited, the tool automatically falls back to EBI BLAST for your homology search. You can also explicitly request BLAST to query specific allowed UniProt or structure databases for comprehensive results.

Can I use a FASTA file for protein sequence similarity search and get Markdown summaries?

Yes, you can use a FASTA file as input for a protein sequence similarity search. The tool interprets alignment metrics like E-value and sequence identity, producing ranked Markdown summaries and optional JSON files for follow-on analysis.

Does this protein homologue search support specific UniProt databases via BLAST?

Yes, the EBI BLAST fallback supports strict allowed database codes. You can explicitly request BLAST to search specific UniProt or structure databases when you need more comprehensive coverage than the default MMseqs2 search provides.

Why does my protein sequence search return no hits or fail to infer function?

A protein sequence search may fail if MMseqs2 is rate-limited and cannot connect, or if the input sequence format is invalid. The tool avoids hallucinated results and raw output parsing, ensuring only verified alignment metrics are reported.