aboutsummaryrefslogtreecommitdiff
path: root/gn3/db
diff options
context:
space:
mode:
Diffstat (limited to 'gn3/db')
-rw-r--r--gn3/db/traits.py212
1 files changed, 146 insertions, 66 deletions
diff --git a/gn3/db/traits.py b/gn3/db/traits.py
index ae1939a..d8d2b62 100644
--- a/gn3/db/traits.py
+++ b/gn3/db/traits.py
@@ -1,5 +1,6 @@
"""This class contains functions relating to trait data manipulation"""
from typing import Any, Dict, Union
+from gn3.function_helpers import compose
def get_trait_csv_sample_data(conn: Any,
@@ -100,119 +101,181 @@ def retrieve_trait_dataset_name(
cursor.execute(query, {"threshold": threshold, "name": name})
return cursor.fetchone()
-PUBLISH_TRAIT_INFO_QUERY = (
- "SELECT "
- "PublishXRef.Id, Publication.PubMed_ID, "
- "Phenotype.Pre_publication_description, "
- "Phenotype.Post_publication_description, "
- "Phenotype.Original_description, "
- "Phenotype.Pre_publication_abbreviation, "
- "Phenotype.Post_publication_abbreviation, "
- "Phenotype.Lab_code, Phenotype.Submitter, Phenotype.Owner, "
- "Phenotype.Authorized_Users, CAST(Publication.Authors AS BINARY), "
- "Publication.Title, Publication.Abstract, Publication.Journal, "
- "Publication.Volume, Publication.Pages, Publication.Month, "
- "Publication.Year, PublishXRef.Sequence, Phenotype.Units, "
- "PublishXRef.comments "
- "FROM "
- "PublishXRef, Publication, Phenotype, PublishFreeze "
- "WHERE "
- "PublishXRef.Id = %(trait_name)s AND "
- "Phenotype.Id = PublishXRef.PhenotypeId AND "
- "Publication.Id = PublishXRef.PublicationId AND "
- "PublishXRef.InbredSetId = PublishFreeze.InbredSetId AND "
- "PublishFreeze.Id =%(trait_dataset_id)s")
-
def retrieve_publish_trait_info(trait_data_source: Dict[str, Any], conn: Any):
"""Retrieve trait information for type `Publish` traits.
https://github.com/genenetwork/genenetwork1/blob/master/web/webqtl/base/webqtlTrait.py#L399-L421"""
+ keys = (
+ "Id", "PubMed_ID", "Pre_publication_description",
+ "Post_publication_description", "Original_description",
+ "Pre_publication_abbreviation", "Post_publication_abbreviation",
+ "Lab_code", "Submitter", "Owner", "Authorized_Users", "Authors",
+ "Title", "Abstract", "Journal", "Volume", "Pages", "Month", "Year",
+ "Sequence", "Units", "comments")
+ columns = (
+ "PublishXRef.Id, Publication.PubMed_ID, "
+ "Phenotype.Pre_publication_description, "
+ "Phenotype.Post_publication_description, "
+ "Phenotype.Original_description, "
+ "Phenotype.Pre_publication_abbreviation, "
+ "Phenotype.Post_publication_abbreviation, "
+ "Phenotype.Lab_code, Phenotype.Submitter, Phenotype.Owner, "
+ "Phenotype.Authorized_Users, CAST(Publication.Authors AS BINARY), "
+ "Publication.Title, Publication.Abstract, Publication.Journal, "
+ "Publication.Volume, Publication.Pages, Publication.Month, "
+ "Publication.Year, PublishXRef.Sequence, Phenotype.Units, "
+ "PublishXRef.comments")
+ query = (
+ "SELECT "
+ "{columns} "
+ "FROM "
+ "PublishXRef, Publication, Phenotype, PublishFreeze "
+ "WHERE "
+ "PublishXRef.Id = %(trait_name)s AND "
+ "Phenotype.Id = PublishXRef.PhenotypeId AND "
+ "Publication.Id = PublishXRef.PublicationId AND "
+ "PublishXRef.InbredSetId = PublishFreeze.InbredSetId AND "
+ "PublishFreeze.Id =%(trait_dataset_id)s").format(columns=columns)
with conn.cursor() as cursor:
cursor.execute(
- PUBLISH_TRAIT_INFO_QUERY,
+ query,
{
k:v for k, v in trait_data_source.items()
if k in ["trait_name", "trait_dataset_id"]
})
- return cursor.fetchone()
+ return dict(zip([k.lower() for k in keys], cursor.fetchone()))
+
+def set_confidential_field(trait_info):
+ """Post processing function for 'Publish' trait types.
-PROBESET_TRAIT_INFO_QUERY = (
- "SELECT "
- "ProbeSet.name, ProbeSet.symbol, ProbeSet.description, "
- "ProbeSet.probe_target_description, ProbeSet.chr, ProbeSet.mb, "
- "ProbeSet.alias, ProbeSet.geneid, ProbeSet.genbankid, ProbeSet.unigeneid, "
- "ProbeSet.omim, ProbeSet.refseq_transcriptid, ProbeSet.blatseq, "
- "ProbeSet.targetseq, ProbeSet.chipid, ProbeSet.comments, "
- "ProbeSet.strand_probe, ProbeSet.strand_gene, "
- "ProbeSet.probe_set_target_region, ProbeSet.proteinid, "
- "ProbeSet.probe_set_specificity, ProbeSet.probe_set_blat_score, "
- "ProbeSet.probe_set_blat_mb_start, ProbeSet.probe_set_blat_mb_end, "
- "ProbeSet.probe_set_strand, ProbeSet.probe_set_note_by_rw, "
- "ProbeSet.flag "
- "FROM "
- "ProbeSet, ProbeSetFreeze, ProbeSetXRef "
- "WHERE "
- "ProbeSetXRef.ProbeSetFreezeId = ProbeSetFreeze.Id AND "
- "ProbeSetXRef.ProbeSetId = ProbeSet.Id AND "
- "ProbeSetFreeze.Name = %(trait_dataset_name)s AND "
- "ProbeSet.Name = %(trait_name)s")
+ It sets the value for the 'confidential' key."""
+ return {
+ **trait_info,
+ "confidential": 1 if (
+ trait_info.get("pre_publication_description", None)
+ and not trait_info.get("pubmed_id", None)) else 0}
def retrieve_probeset_trait_info(trait_data_source: Dict[str, Any], conn: Any):
"""Retrieve trait information for type `ProbeSet` traits.
https://github.com/genenetwork/genenetwork1/blob/master/web/webqtl/base/webqtlTrait.py#L424-L435"""
+ keys = (
+ "name", "symbol", "description", "probe_target_description", "chr",
+ "mb", "alias", "geneid", "genbankid", "unigeneid", "omim",
+ "refseq_transcriptid", "blatseq", "targetseq", "chipid", "comments",
+ "strand_probe", "strand_gene", "probe_set_target_region", "proteinid",
+ "probe_set_specificity", "probe_set_blat_score",
+ "probe_set_blat_mb_start", "probe_set_blat_mb_end", "probe_set_strand",
+ "probe_set_note_by_rw", "flag")
+ query = (
+ "SELECT "
+ "{columns} "
+ "FROM "
+ "ProbeSet, ProbeSetFreeze, ProbeSetXRef "
+ "WHERE "
+ "ProbeSetXRef.ProbeSetFreezeId = ProbeSetFreeze.Id AND "
+ "ProbeSetXRef.ProbeSetId = ProbeSet.Id AND "
+ "ProbeSetFreeze.Name = %(trait_dataset_name)s AND "
+ "ProbeSet.Name = %(trait_name)s").format(
+ columns=", ".join(["ProbeSet.{}".format(x) for x in keys]))
with conn.cursor() as cursor:
cursor.execute(
- PROBESET_TRAIT_INFO_QUERY,
+ query,
{
k:v for k, v in trait_data_source.items()
if k in ["trait_name", "trait_dataset_name"]
})
- return cursor.fetchone()
-
-GENO_TRAIT_INFO_QUERY = (
- "SELECT "
- "Geno.name, Geno.chr, Geno.mb, Geno.source2, Geno.sequence "
- "FROM "
- "Geno, GenoFreeze, GenoXRef "
- "WHERE "
- "GenoXRef.GenoFreezeId = GenoFreeze.Id AND GenoXRef.GenoId = Geno.Id AND "
- "GenoFreeze.Name = %(trait_dataset_name)s AND Geno.Name = %(trait_name)s")
+ return dict(zip(keys, cursor.fetchone()))
def retrieve_geno_trait_info(trait_data_source: Dict[str, Any], conn: Any):
"""Retrieve trait information for type `Geno` traits.
https://github.com/genenetwork/genenetwork1/blob/master/web/webqtl/base/webqtlTrait.py#L438-L449"""
+ keys = ("name", "chr", "mb", "source2", "sequence")
+ query = (
+ "SELECT "
+ "{columns} "
+ "FROM "
+ "Geno, GenoFreeze, GenoXRef "
+ "WHERE "
+ "GenoXRef.GenoFreezeId = GenoFreeze.Id AND GenoXRef.GenoId = Geno.Id AND "
+ "GenoFreeze.Name = %(trait_dataset_name)s AND "
+ "Geno.Name = %(trait_name)s").format(
+ columns=", ".join(["Geno.{}".format(x) for x in keys]))
with conn.cursor() as cursor:
cursor.execute(
- GENO_TRAIT_INFO_QUERY,
+ query,
{
k:v for k, v in trait_data_source.items()
if k in ["trait_name", "trait_dataset_name"]
})
- return cursor.fetchone()
-
-TEMP_TRAIT_INFO_QUERY = (
- "SELECT name, description FROM Temp "
- "WHERE Name = %(trait_name)s")
+ return dict(zip(keys, cursor.fetchone()))
def retrieve_temp_trait_info(trait_data_source: Dict[str, Any], conn: Any):
"""Retrieve trait information for type `Temp` traits.
https://github.com/genenetwork/genenetwork1/blob/master/web/webqtl/base/webqtlTrait.py#L450-452"""
+ keys = ("name", "description")
+ query = (
+ "SELECT {columns} FROM Temp "
+ "WHERE Name = %(trait_name)s").format(columns=", ".join(keys))
with conn.cursor() as cursor:
cursor.execute(
- TEMP_TRAIT_INFO_QUERY,
+ query,
{
k:v for k, v in trait_data_source.items()
if k in ["trait_name"]
})
- return cursor.fetchone()
+ return dict(zip(keys, cursor.fetchone()))
+
+def set_haveinfo_field(trait_info):
+ """
+ Common postprocessing function for all trait types.
+
+ Sets the value for the 'haveinfo' field."""
+ return {**trait_info, "haveinfo": 1 if trait_info else 0}
+
+def set_homologene_id_field_probeset(trait_info, conn):
+ """
+ Postprocessing function for 'ProbeSet' traits.
+
+ Sets the value for the 'homologene' key.
+ """
+ query = (
+ "SELECT HomologeneId FROM Homologene, Species, InbredSet"
+ " WHERE Homologene.GeneId = %(geneid)s AND InbredSet.Name = %(riset)s"
+ " AND InbredSet.SpeciesId = Species.Id AND"
+ " Species.TaxonomyId = Homologene.TaxonomyId")
+ with conn.cursor() as cursor:
+ cursor.execute(
+ query,
+ {
+ k:v for k, v in trait_info.items()
+ if k in ["geneid", "riset"]
+ })
+ res = cursor.fetchone()
+ if res:
+ return {**trait_info, "homologeneid": res[0]}
+ return {**trait_info, "homologeneid": None}
+
+def set_homologene_id_field(trait_info, conn):
+ """
+ Common postprocessing function for all trait types.
+
+ Sets the value for the 'homologene' key."""
+ set_to_null = lambda ti: {**ti, "homologeneid": None}
+ functions_table = {
+ "Temp": set_to_null,
+ "Geno": set_to_null,
+ "Publish": set_to_null,
+ "ProbeSet": lambda ti: set_homologene_id_field_probeset(ti, conn)
+ }
+ return functions_table[trait_info["type"]](trait_info)
def retrieve_trait_info(
trait_type: str, trait_name: str, trait_dataset_id: int,
- trait_dataset_name: str, conn: Any):
+ trait_dataset_name: str, conn: Any, QTL=None):
"""Retrieves the trait information.
https://github.com/genenetwork/genenetwork1/blob/master/web/webqtl/base/webqtlTrait.py#L397-L456
@@ -225,7 +288,24 @@ def retrieve_trait_info(
"Geno": retrieve_geno_trait_info,
"Temp": retrieve_temp_trait_info
}
- return trait_info_function_table[trait_type](
+
+ common_post_processing_fn = compose(
+ lambda ti: set_homologene_id_field(ti, conn),
+ lambda ti: {"type": trait_type, **ti},
+ set_haveinfo_field)
+
+ trait_post_processing_functions_table = {
+ "Publish": compose(set_confidential_field, common_post_processing_fn),
+ "ProbeSet": compose(common_post_processing_fn),
+ "Geno": common_post_processing_fn,
+ "Temp": common_post_processing_fn
+ }
+
+ retrieve_info = compose(
+ trait_post_processing_functions_table[trait_type],
+ trait_info_function_table[trait_type])
+
+ return retrieve_info(
{
"trait_name": trait_name,
"trait_dataset_id": trait_dataset_id,
generated by cgit v1.2.3 (git 2.46.2) at 2024-12-04 17:10:13 +0000