Add similarity-search-parallel: thread-pool scoring, byte-identical to similarity-search
coordinator / test (push) Canceled after 0s
python / test (push) Canceled after 0s
release / binaries (amd64, darwin) (push) Canceled after 0s
release / binaries (amd64, linux) (push) Canceled after 0s
release / binaries (amd64, windows) (push) Canceled after 0s
release / binaries (arm64, darwin) (push) Canceled after 0s
release / binaries (arm64, linux) (push) Canceled after 0s
release / binaries (arm64, windows) (push) Canceled after 0s
release / wheel (push) Canceled after 0s
release / image (push) Canceled after 0s
users / test (push) Canceled after 0s
release / release (push) Canceled after 0s

This commit is contained in:
Emil
2026-08-03 15:12:18 +03:00
parent b63441bb8c
commit 5b2d5b5f6e
9 changed files with 771 additions and 2 deletions
@@ -597,6 +597,206 @@
"upload_ready": true,
"verifier": "exact-artifact@1",
"version": "1.0.0"
},
{
"capabilities": [
"similarity-search-parallel"
],
"description": "Exact top-k Tanimoto molecular similarity search over deterministic TSV shards with a bounded merge; each shard is fingerprinted and scored across a thread pool. Output is byte-identical to similarity-search.",
"determinism": "byte_exact",
"enabled": true,
"inputs": {
"input": {
"allow_nested_collections": false,
"canonicalizer": "scimesh-tsv-v1",
"encoding": "utf-8",
"max_bytes": 10737418240,
"max_dimensions": [],
"max_records": 100000000,
"media_type": "text/tab-separated-values",
"privacy_class": "project",
"ref": "molecule-table@1",
"retention_class": "durable",
"streaming": false,
"validator": "delimited-table@1",
"validator_configuration": {
"required_columns": [
"canonical_smiles",
"chembl_id"
]
}
}
},
"name": "similarity-search-parallel",
"outputs": {
"result": {
"allow_nested_collections": false,
"canonicalizer": "scimesh-search-result-v1",
"encoding": "utf-8",
"max_bytes": 1073741824,
"max_dimensions": [],
"max_records": 100000,
"media_type": "text/csv",
"privacy_class": "project",
"ref": "similarity-search-result@1",
"retention_class": "durable",
"streaming": false,
"validator": "delimited-table@1",
"validator_configuration": {
"columns": [
"rank",
"chembl_id",
"canonical_smiles",
"similarity"
]
}
}
},
"parameters_schema": {
"additionalProperties": false,
"oneOf": [
{
"not": {
"required": [
"query_smiles"
]
},
"required": [
"query_id"
]
},
{
"not": {
"required": [
"query_id"
]
},
"required": [
"query_smiles"
]
}
],
"properties": {
"max_rows": {
"minimum": 1,
"type": "integer"
},
"progress_every": {
"minimum": 0,
"type": "integer"
},
"query_id": {
"maxLength": 200,
"minLength": 1,
"type": "string"
},
"query_smiles": {
"maxLength": 200,
"minLength": 1,
"type": "string"
},
"threads": {
"description": "Threads used to fingerprint and score one shard (default: CPU count).",
"minimum": 1,
"type": "integer"
},
"threshold": {
"maximum": 1,
"minimum": 0,
"type": "number"
},
"threshold_direction": {
"enum": [
"greater",
"less"
]
},
"top_k": {
"minimum": 1,
"type": "integer"
}
},
"type": "object"
},
"reduction": "top-k",
"trust_modes": [
"trusted",
"untrusted_quorum"
],
"ui_elements": [
{
"default": null,
"field": "query_id",
"group": "",
"help": "ChEMBL id of the query molecule. Provide exactly one of id or SMILES.",
"label": "Query molecule id",
"options": [],
"order": 1,
"placeholder": "",
"widget": "text"
},
{
"default": null,
"field": "query_smiles",
"group": "",
"help": "SMILES of the query molecule. Provide exactly one of id or SMILES.",
"label": "Query molecule SMILES",
"options": [],
"order": 2,
"placeholder": "",
"widget": "text"
},
{
"default": 20,
"field": "top_k",
"group": "",
"help": "Number of most similar molecules to keep per shard (global merge keeps the best of these).",
"label": "Top k",
"options": [],
"order": 3,
"placeholder": "",
"widget": "number"
},
{
"default": "greater",
"field": "threshold_direction",
"group": "",
"help": "Keep molecules with similarity greater or less than the threshold.",
"label": "Direction",
"options": [
"greater",
"less"
],
"order": 4,
"placeholder": "",
"widget": "select"
},
{
"default": null,
"field": "threshold",
"group": "",
"help": "Optional similarity bound: results are filtered to this direction.",
"label": "Similarity threshold",
"options": [],
"order": 5,
"placeholder": "e.g. 0.8",
"widget": "number"
},
{
"default": null,
"field": "threads",
"group": "",
"help": "Threads used to fingerprint and score one shard (default: CPU count).",
"label": "Threads per shard",
"options": [],
"order": 6,
"placeholder": "auto",
"widget": "number"
}
],
"upload_ready": true,
"verifier": "exact-artifact@1",
"version": "1.0.0"
}
]
}