Skip to main content
Glama

hmm_search

Read-only

Score sequences against ONE profile HMM on this machine, as hmmsearch scores a family against a set: per sequence a Forward and a Viterbi bit score, an E-value when the model is calibrated, every domain with its model and sequence span and its three-line alignment, and the family's own gathering-cutoff verdict when the model publishes one (Pfam does).

Tool call template

This tool declares no required parameters. Add optional parameters from the input schema as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "hmm_search"
  }
}

Input Schema

NameRequiredDescriptionDefault
ZNothe database size each E-value is scaled by. 0 takes the number of sequences screened, which is what hmmsearch does with the database it searched; set it to compare against a larger search.
modeNo`local` finds the model anywhere in a sequence, as hmmsearch does: right for a domain inside a longer protein. `glocal` makes the whole model align, which a full-length member of the family can and a fragment cannot. Values: local (Local (hmmsearch)), glocal (Glocal (whole model)).local
pfamNoa Pfam accession (PF and five digits): the family’s current HMM is fetched from InterPro for this run. Its gathering cutoff comes with it.
_filesNoThe project: record references shaped like _subject. Their records become library documents a tool names in sequences/parts/document; a trace or alignment is found by file name.
profileNothe `.hmm` file to score against, by name, from the project or the `_files` sent with the call. Left empty with no `profile_text` or `pfam`, the one `.hmm` present is used, and several are refused by name.
_subjectNoThe record to act on: exactly one of sequence, accession (+source), sample (an id from list_samples) or file ({name, text} or {name, base64}); record_index picks one record of several.
alphabetNowhat the sequences are. `auto` reads them the way HMMER does (Easel’s composition rule). Set it when a short or unusual set is read wrongly: every DNA base is also an amino-acid letter, so a nucleotide set against a protein model scores without complaint and means nothing. Values: auto (Auto (Easel’s rule)), protein (Protein), nucleotide (Nucleotide). Advanced.auto
multihitNoallow several separate domains in one sequence (the J state). Off allows one: right when a family occurs once per protein.
sequencesNothe sequences to screen: wire upstream records or sequences, or list library documents. A multi-record file named whole contributes every record. Gaps are removed before scoring.
profile_textNothe model itself as HMMER3 text; wire `hmm_build`’s `hmm` here to screen with a model you just built. Advanced.
include_subjectNoalso screen the open or wired record (a protein record by its residues).

Output Schema

NameRequiredDescriptionDefault
hitsYesone row per sequence, best Forward score first: name, length, forwardBits, viterbiBits, forwardEvalue (absent for an uncalibrated model), membership against the gathering cutoff, and each domain’s seqFrom/seqTo (1-based) and hmmFrom/hmmTo with its alignment
countYessequences screened
membersYessequences at or above the family’s gathering cutoff (0 when the model publishes none)
profileYesthe model screened against: name, accession, match states, alphabet, cutoffs, whether it carries E-value statistics

Schema Changelog

Changes observed during successful MCP inspections.

  1. Added

TDQS

A3.6/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already declare readOnlyHint=true and openWorldHint=false, but the description adds genuine behavioral nuance the annotations cannot: E-values exist only when the model is calibrated, and a gathering-cutoff verdict only when the model publishes one. 'On this machine' reinforces the local, non-open-world nature; runtime, size limits and failure modes remain unstated.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

One sentence, front-loaded with the action and scope, then a dense enumeration of what comes back. Every clause carries signal and nothing is repeated or padded.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

With 11 optional parameters, 100% schema coverage and an output schema present, the description need not restate returns in full, yet it usefully previews them. The one gap is the absence of any when-to-use framing against the many sibling search tools, which for an 11-parameter tool leaves a small hole.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so every parameter (Z, mode, pfam, profile, multihit, alphabet) is already documented with defaults and enum meanings. The description adds no further parameter meaning beyond the schema, so the baseline of 3 applies.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

A specific verb and resource: 'Score sequences against ONE profile HMM on this machine,' which plainly sets it apart from model-building or generic sequence search. The capitalization of 'ONE' usefully signals the single-model scope, though no sibling (e.g. hmm_build, blast_search) is named to anchor the contrast.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description never says when to reach for this rather than blast_search or another search sibling, nor when not to. The only usage routing ('wire `hmm_build`'s `hmm` here') lives in the parameter schema, not the description, leaving the description itself without guidance.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources