> This page is for MSA Search, version 2.0.0.
> For other versions, use one of these documentation indexes:
> - 2.6.0 (Latest) (default): https://docs.nvidia.com/nim/bionemo/msa-search/2.6.0/llms.txt
> - 2.5.0: https://docs.nvidia.com/nim/bionemo/msa-search/2.5.0/llms.txt
> - 2.4.0: https://docs.nvidia.com/nim/bionemo/msa-search/2.4.0/llms.txt
> - 2.3.0: https://docs.nvidia.com/nim/bionemo/msa-search/2.3.0/llms.txt
> - 2.2.0: https://docs.nvidia.com/nim/bionemo/msa-search/2.2.0/llms.txt
> - 2.1.0: https://docs.nvidia.com/nim/bionemo/msa-search/2.1.0/llms.txt
> - 2.0.0: https://docs.nvidia.com/nim/bionemo/msa-search/2.0.0/llms.txt
> - 1.0.0: https://docs.nvidia.com/nim/bionemo/msa-search/1.0.0/llms.txt

> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/nim/bionemo/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/nim/bionemo/_mcp/server.

# API Reference

This documentation contains the API reference for the MSA Search NIM.

## OpenAPI Specification

You can download or view the OpenAPI specification when the NIM is running:

```bash
curl http://localhost:8000/openapi.json
```

You can also navigate to the interactive API documentation at `http://localhost:8000/docs` in your browser.

## Multiple Sequence Alignment Search

Endpoint path: `/biology/colabfold/msa-search/predict`

Request type: `POST`

### Input Parameters

- `sequence` (string, **required**): A sequence to search against the MSA databases. Must be a valid protein sequence composed of the 20 standard amino acids (ARNDCQEGHILKMFPSTWYV). Length: 1-4096 characters.

  Example:
  ```json
  "SGSMKTAISLPDETFDRVSRRASELGMSRSEFFTKAAQR"
  ```

- `databases` (list[string], optional): A comma-separated list of database names, to search against (all databases are searched, by default). Accepts 1-3 database names. Default: `["all"]`.

  Examples: `["all"]`, `["Uniref30_2302"]`, `["Uniref30_2302", "PDB70_220313"]`

- `search_type` (string, optional): Which type of MSA Search to run for alignment production. Default: `"colabfold"`.

  Options:
  - `"colabfold"`: Cascaded search with higher sensitivity
  - `"alphafold2"`: Single-pass iterative search

  Examples: `"colabfold"`, `"alphafold2"`

- `e_value` (float, optional): The e-value threshold for filtering hits when building the Multiple Sequence Alignment. Sequences with an e-value greater than this are not included in the MSA. Range: 0.0-1.0. Default: `0.0001`.

- `iterations` (int, optional): The number of MSA iterations to perform, where more iterations find more distant homologs. Range: 1-6. Default: `1`. **Note**: For cascaded search (`search_type="colabfold"`), the number of iterations is fixed to 3 and this parameter is ignored.

- `max_msa_sequences` (int, optional): The maximum sequences taken from the MSA for model prediction. Range: 1-10001. Default: `500`. **Note**: When GPU Server is enabled (default in version 2.0.0), this parameter must be set globally via the `NIM_GLOBAL_MAX_MSA_DEPTH` environment variable at container startup.

- `output_alignment_formats` (list[string], optional): The output format of the MSA. Supported formats: `"a3m"`, `"fasta"`. Default: `["a3m"]`.

  Examples: `["a3m"]`, `["a3m", "fasta"]`

### Outputs

- `alignments` (Dictionary[string → Dictionary[string → AlignmentFileRecord]]): The MSA alignments organized by database and format. Structure: `{database_name: {format: AlignmentFileRecord}}`. For example, `alignments['Uniref30_2302']['a3m']` contains the A3M alignment for the Uniref30 database.

  Each `AlignmentFileRecord` contains:
  - `alignment` (string): The contents of a single MSA. For 'a3m' format, sequences are in compact A3M format with insertions in lowercase. For 'fasta' format, sequences are in standard aligned FASTA format with gaps.
  - `format` (string): The format of the alignment record(s). Values: `"a3m"` or `"fasta"`.

  Example response:
  ```json
  {
    "alignments": {
      "Uniref30_2302": {
        "a3m": {
          "alignment": ">query\nMVPSAGQLALFALGIVLAACQALENS\n>hit1\nMVPSAGQLALFALGIV---CQALENS\n>hit2\nMVPSAGQLALF-LGIV---CQALENS",
          "format": "a3m"
        }
      }
    },
    "metrics": {}
  }
  ```

  To access the A3M-formatted alignment for the Uniref30_2302 database:
  ```python
  alignments["Uniref30_2302"]["a3m"]["alignment"]
  ```

- `metrics` (dictionary, optional): Contains information about the response useful for debugging and measuring performance. May be empty or null.

## Get Database Configuration

Endpoint path: `/biology/colabfold/msa-search/config/msa-database-configs`

Request type: `GET`

### Input Parameters

None.

### Outputs

- `configs` (dictionary): A nested dictionary containing information about all configured MSA databases. The configuration can be converted to YAML format to view the original database configuration.

## Get MMSeqs2 Version

Endpoint path: `/biology/colabfold/msa-search/mmseqs2/version`

Request type: `GET`

### Input Parameters

None.

### Outputs

Returns version information for the MMSeqs2 installation used by the NIM.

<Note>
Use this endpoint to get the exact MMSeqs2 version when you need to create custom database indices. Custom indices must be created with the same MMSeqs2 version as the one running in the NIM to ensure compatibility.
</Note>

## Health Endpoints

### Readiness Check

Endpoint path: `/v1/health/ready`

Request type: `GET`

**Description**: Checks if the service is ready to handle requests.

#### Outputs

- Status code `200`: Service is ready
- Status code `503`: Service is not ready

Response includes a JSON object with:
- `message` (string): Status message
- `object` (string): Always "health.response"
- `status` (string, optional): Status string for backwards compatibility

### Liveness Check

Endpoint path: `/v1/health/live`

Request type: `GET`

**Description**: Checks if the service is live (running).

#### Outputs

- Status code `200`: Service is live
- Status code `503`: Service is not live

Response format is the same as the readiness check.

## NIM Metadata Endpoints

### Version

Endpoint path: `/v1/version`

Request type: `GET`

**Description**: Returns version information for the NIM.

#### Outputs

- `release` (string): The product release version of the NIM
- `api` (string): The server API version running inside the NIM

### License

Endpoint path: `/v1/license`

Request type: `GET`

**Description**: Returns license information for the NIM.

#### Outputs

- `name` (string): The name of the license
- `path` (string): The filepath within the container containing the license content
- `sha` (string): SHA1 hash of the license contents
- `size` (integer): Number of characters in the license content
- `url` (string): URL where the license is hosted externally
- `type` (string): Always "file"
- `content` (string): The full license text

### Metadata

Endpoint path: `/v1/metadata`

Request type: `GET`

**Description**: Returns comprehensive metadata about the NIM deployment.

#### Outputs

- `assetInfo` (list[string]): Required container assets excluding model artifacts
- `licenseInfo` (LicenseEndpointModel): License information
- `modelInfo` (list[ModelInfo]): Information about models being served
- `repository_override` (string): Alternate location for retrieving artifacts
- `version` (string): NIM service version
- `selectedModelProfileId` (string): ID of the currently selected model profile

### Manifest

Endpoint path: `/v1/manifest`

Request type: `GET`

**Description**: Returns the manifest file describing required model artifacts.

#### Outputs

- `manifest_file` (string): Content of the manifest file
- `repository_override` (string): Alternate location for retrieving artifacts

### Metrics

Endpoint path: `/v1/metrics`

Request type: `GET`

**Description**: Exposes Prometheus metrics for monitoring.

#### Outputs

Returns metrics in Prometheus format.