Protein Sequence Similarity Search by google-deepmind — AI Skill for Any LLM

Overview

Publisher	google-deepmind
Repository	`science-skills`
Skill name	`protein-sequence-similarity-search`
Stars	3.1K
Forks	349
Bundled files	3
License	Apache-2.0
Links	Skill source Repository Homepage

Markdown instructions

A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

Works with any LLM

AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

3 bundled files

Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

Open source

Published by google-deepmind on GitHub. Read the source before you install it.

Installation

Install the Protein Sequence Similarity Search AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

Open the app and go to Plugins → Skills.
Choose "Install from GitHub".
Paste the skill folder URL below and confirm.
Enable the skill in any chat where you want it available.

Store

Installed

Skills

Add skills

Import one or more skills from a public GitHub folder or a local .skill or .zip archive, or SKILL.md file.

From GitHub URLFrom .skill, .zip, or SKILL.md

GitHub repo or folder

https://github.com/google-deepmind/science-skills/tree/main/skills/protein_sequence_similarity_search

Points to a single skill folder, or any folder that contains multiple skills. Public repos only.

CancelContinue

Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills

git clone --depth 1 https://github.com/google-deepmind/science-skills.git /tmp/science-skills
mkdir -p .claude/skills
cp -r /tmp/science-skills/skills/protein_sequence_similarity_search .claude/skills/protein-sequence-similarity-search

Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Protein Sequence Similarity Search in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Protein Sequence Similarity Search on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Protein Sequence Similarity Search is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

Loaded only when it is needed

The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

Switch models mid-chat

Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Prerequisites

uv: Read the uv skill and follow its Setup instructions to ensure uv is installed and on PATH.
User Notification: If .licenses/protein_sequence_similarity_search_LICENSE.txt does not already exist in the workspace root directory then (1) prominently notify the user to check the terms at https://www.ebi.ac.uk/jdispatcher/sss/ncbiblast and https://colabfold.com, then (2) create the file recording the notification text and timestamp.
.env file: Make sure the .env file exists in your home directory. Create one if it does not exist.
USER_EMAIL (optional but recommended): Recommended by the EBI for BLAST job tracking, but the skill works without it. You MUST use the safe credentials protocol in the credentials skill to check for and request this credential if this skill looks relevant to the user's request.

Goal

Take a user-provided amino acid sequence (or a path to a .fasta file), search for sequence homologues using the fastest available method, generate a Markdown-formatted table of the top hits, interpret key alignment metrics, summarize the inferred protein functions, and save results locally for future programmatic analysis.

Core Rules

Strict Validation: For BLAST, only use database codes listed in the table below.
No Hallucinations: If a script throws an error or returns no hits, inform the user clearly. Do NOT invent sequence homologues.
Do Not Parse Output Files: Do not parse the JSON, a3m, or any other raw output files. Rely on the generated .md file for your summary. The JSON and other outputs are for subsequent tool use only.
Always State the Method: Every report must clearly state whether the search used the quick MMseqs2 (ColabFold API) or the slower EBI BLAST method.
Notification: If this skill is used, ensure this is mentioned in the output. Explicitly state that the corresponding program (MMSEQS2 or EBI BLAST) and Sequence Databases were used.

Search Method Selection

Choose the search method based on the user's request:

If the user says "quick search" or "fast search", no specific method requested / general homologue search, of if you are unsure: Run MMseqs2 (fast, default) using mmseqs2_search.py

If MMseqs2 fails (exit code 2: RATELIMIT or API error) or User explicitly requests "BLAST" or a specific BLAST database (e.g. uniprotkb_swissprot, pdb, uniprotkb_human): Run BLAST using uniprot_blast.py

Instructions

Identify the query from the user. It can be a raw sequence string (e.g., "MKVLY...") or a path to a local file (e.g., "./data/sequence.fasta").
Determine the search method using the list above.

Path A: MMseqs2 Search (Default)

Generate File Names: Generate descriptive output file names based on the input (e.g., proteinA_mmseqs2.json and proteinA_mmseqs2.md).

Execute the MMseqs2 script:

Default:


uv run scripts/mmseqs2_search.py <SEQUENCE_OR_FILE> -o <generated-filename.md> -j <generated-filename.json>

With mgnify:


uv run scripts/mmseqs2_search.py <SEQUENCE_OR_FILE> -o <generated-filename.md> -j <generated-filename.json> --include-mgnify

The script will query the ColabFold MMseqs2 API and poll for completion. This is typically fast (under 2 minutes).
If the script exits with code 2 (API failure, rate limit), automatically fall back to BLAST (Path B below). Inform the user: "MMseqs2 search failed, falling back to BLAST."
Read the Results: Open and read the generated .md file.

Path B: BLAST Search (Explicit or Fallback)

Database Selection & Validation: Determine the most appropriate database(s) based on the user's prompt.
- Consult the Available BLAST Databases table below.
- If the user specifies a taxonomic group (e.g., "Find homologues in microbes"), select the corresponding Database Code (e.g., uniprotkb_bacteria).
- If the user explicitly requests curated hits, use uniprotkb_swissprot.
- If no specific database is requested, do not specify --databases.
- Validation: Ensure the database code exactly matches an entry in the table. If the user requests a database not on the list, do not proceed and provide the allowed list.
Generate File Names: (e.g., proteinA_ebi_blast.json and proteinA_ebi_blast.md).
This API requires the user email address to be set in the USER_EMAIL environment variable for inclusion in request header. You MUST use the safe credentials protocol in the credentials skill to check for and request this credential if this skill looks relevant to the user's request.

Execute the BLAST script:

Default (uniprotkb):


uv run scripts/uniprot_blast.py <SEQUENCE_OR_FILE> -o <generated-filename.md> -j <generated-filename.json>

Custom database:


uv run scripts/uniprot_blast.py <SEQUENCE_OR_FILE> -o <generated-filename.md> -j <generated-filename.json> --databases <db1,db2>

The script will query the EBI BLAST API and poll the server. Note: This can take up to 15 minutes; wait patiently.
Read the Results: Open and read the generated .md file.

Common Steps (Both Methods)

Interpret the Metrics: Summarize the top 3 to 5 sequence homologues. Assess match quality using:
- Q-Cov (Query Coverage): High percentages mean the match covers most of the query sequence.
- E-value: Lower E-values (e.g., 1e-50) indicate extreme statistical significance.
- Seq Identity: Provides evolutionary context (highly conserved vs. distant homologue).
Perform Functional Analysis:
- If the results table includes protein descriptions, analyze them directly: report specific protein names/functions of the top homologues and summarize the variety of functions, domains, or protein families found.
- If the results contain only UniProt accession IDs without descriptions (common with MMseqs2), look up the protein names and functions for the top 3–5 hits using the uniprot-database skill or other appropriate methods before summarizing.
Inform the user of both newly created files (.json and .md) and their locations.

Available BLAST Databases

uniprotkb – UniProt Knowledgebase (The UniProt Knowledgebase includes UniProtKB/Swiss-Prot and UniProtKB/TrEMBL): The UniProt Knowledgebase (UniProtKB) is the central access point for extensive curated protein information, including function, classification, and cross-references. Search UniProtKB to retrieve "everything that is known" about a particular sequence
uniprotkb_swissprot – UniProtKB/Swiss-Prot (The manually annotated section of UniProtKB): The manually curated subsection of the UniProt Knowledgebase
uniprotkb_swissprotsv – UniProtKB/Swiss-Prot isoforms (The manually annotated isoforms of UniProtKB/Swiss-Prot): The isoform sequences for the manually curated subsection of the UniProt Knowledgebase
uniprotkb_reference_proteomes – UniProtKB Reference Proteomes: Taxonomic subset of the UniProtKB Reference Proteomes
uniprotkb_trembl – UniProtKB/TrEMBL (The automatically annotated section of UniProtKB): Subsection of the UniProt Knowledgebase derived from ENA Sequence (formerly EMBL-Bank) coding sequence translations with annotation produced by an automated process
uniprotkb_refprotswissprot – UniProtKB Reference Proteomes plus Swiss-Prot: UniProtKB Reference Proteomes plus Swiss-Prot
uniprotkb_archaea – UniProtKB Archaea: Taxonomic subset of the UniProt Knowledgebase for archaea
uniprotkb_arthropoda – UniProtKB Arthropoda: Taxonomic subset of the UniProt Knowledgebase for arthropoda
uniprotkb_bacteria – UniProtKB Bacteria: Taxonomic subset of the UniProt Knowledgebase for bacteria
uniprotkb_complete_microbial_proteomes – UniProtKB Complete Microbial Proteomes: Taxonomic subset of the UniProt Knowledgebase for complete microbial proteomes
uniprotkb_eukaryota – UniProtKB Eukaryota: Taxonomic subset of the UniProt Knowledgebase for eukaryota
uniprotkb_fungi – UniProtKB Fungi: Taxonomic subset of the UniProt Knowledgebase for fungi
uniprotkb_human – UniProtKB Human: Taxonomic subset of the UniProt Knowledgebase for human
uniprotkb_mammals – UniProtKB Mammals: Taxonomic subset of the UniProt Knowledgebase for mammals
uniprotkb_nematoda – UniProtKB Nematoda: Taxonomic subset of the UniProt Knowledgebase for nematoda
uniprotkb_rodents – UniProtKB Rodents: Taxonomic subset of the UniProt Knowledgebase for rodents
uniprotkb_vertebrates – UniProtKB Vertebrates: Taxonomic subset of the UniProt Knowledgebase for vertebrates
uniprotkb_viridiplantae – UniProtKB Viridiplantae: Taxonomic subset of the UniProt Knowledgebase for viridiplantae
uniprotkb_viruses – UniProtKB Viruses: Taxonomic subset of the UniProt Knowledgebase for viruses
uniprotkb_enzyme – UniProtKB Enzyme: Taxonomic subset of the UniProt Knowledgebase for enzymes
uniprotkb_covid19 – UniProtKB COVID-19: Taxonomic subset of the UniProt Knowledgebase for COVID-19
uniref100 – UniProt Clusters 100% (UniRef100): The UniProt Reference Clusters (UniRef) containing sequences which are 100% identical.
uniref90 – UniProt Clusters 90% (UniRef90): The UniProt Reference Clusters (UniRef) containing sequences which are 90% identical.
uniref50 – UniProt Clusters 50% (UniRef50): The UniProt Reference Clusters (UniRef) containing sequences which are 50% identical.
pdb – Protein Structure Sequences (PDBe protein structure sequences): Protein sequences from structures described in the Brookhaven Protein Data Bank (PDB)

Frequently asked questions

What does the Protein Sequence Similarity Search AI skill do?

Searches for homologous protein sequences using MMseqs2 (fast, default) or BLAST (comprehensive, fallback). Trigger this whenever the user provides a protein sequence or FASTA file and asks to find homologues, sequence matches, or wants to infer protein function based on sequence similarity, but not when the user wants to infer protein function based on structural similarity.

Why use Protein Sequence Similarity Search on TypingMind?

Because you install it once and use it with any model. Protein Sequence Similarity Search is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Protein Sequence Similarity Search in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/google-deepmind/science-skills/tree/main/skills/protein_sequence_similarity_search. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Protein Sequence Similarity Search?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Protein Sequence Similarity Search?

As many as you like. As long as a model supports skills, you can use Protein Sequence Similarity Search with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Protein Sequence Similarity Search AI skill free?

Yes. It is published on GitHub by google-deepmind under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

Overview

Installation

Install in TypingMind

Install in another agent

Use it in TypingMind

Works with any AI model

Skill instructions

Prerequisites

Goal

Core Rules

Search Method Selection

Instructions

Path A: MMseqs2 Search (Default)

Path B: BLAST Search (Explicit or Fallback)

Common Steps (Both Methods)

Available BLAST Databases

Bundled files

Frequently asked questions

Related AI Skills

Set up your own AI workspace now