aboutsummaryrefslogtreecommitdiff
path: root/gn3
diff options
context:
space:
mode:
authorMuriithi Frederick Muriuki2021-08-09 14:22:54 +0300
committerMuriithi Frederick Muriuki2021-08-09 14:22:54 +0300
commit8f022ae1a31224d0526443ad9779f30206b4a770 (patch)
tree672349afe35b643a3834465f896102a8f0d43d4e /gn3
parent9ed71447892922c0039dd749a9052be502f1b818 (diff)
downloadgenenetwork3-8f022ae1a31224d0526443ad9779f30206b4a770.tar.gz
Retrieve the trait data
Issue: https://github.com/genenetwork/gn-gemtext-threads/blob/main/topics/gn1-migration-to-gn2/clustering.gmi * Add functions to retrieve the `value`, `variance`, and `ndata` values for any given trait.
Diffstat (limited to 'gn3')
-rw-r--r--gn3/db/traits.py245
1 files changed, 244 insertions, 1 deletions
diff --git a/gn3/db/traits.py b/gn3/db/traits.py
index be46437..a740352 100644
--- a/gn3/db/traits.py
+++ b/gn3/db/traits.py
@@ -1,5 +1,5 @@
"""This class contains functions relating to trait data manipulation"""
-from typing import Any, Dict, Union
+from typing import Any, Dict, Union, Sequence
from gn3.function_helpers import compose
from gn3.db.datasets import retrieve_trait_dataset
@@ -408,3 +408,246 @@ def retrieve_trait_info(
"riset": trait_dataset["riset"]
}
return trait_info
+
+def retrieve_temp_trait_data(trait_info: dict, conn: Any):
+ """
+ Retrieve trait data for `Temp` traits.
+ """
+ query = (
+ "SELECT "
+ "Strain.Name, TempData.value, TempData.SE, TempData.NStrain, "
+ "TempData.Id "
+ "FROM TempData, Temp, Strain "
+ "WHERE TempData.StrainId = Strain.Id "
+ "AND TempData.Id = Temp.DataId "
+ "AND Temp.name = %(trait_name)s "
+ "ORDER BY Strain.Name")
+ with conn.cursor() as cursor:
+ cursor.execute(
+ query,
+ {"trait_name": trait_info["trait_name"]})
+ return [dict(zip(
+ ["strain_name", "value", "se_error", "nstrain", "id"], row))
+ for row in cursor.fetchall()]
+ return []
+
+def retrieve_species_id(riset, conn: Any):
+ """
+ Retrieve a species id given the RISet value
+ """
+ with conn.cursor as cursor:
+ cursor.execute(
+ "SELECT SpeciesId from InbredSet WHERE Name = %(riset)s",
+ {"riset": riset})
+ return cursor.fetchone()[0]
+ return None
+
+def retrieve_geno_trait_data(trait_info: Dict, conn: Any):
+ """
+ Retrieve trait data for `Geno` traits.
+ """
+ query = (
+ "SELECT Strain.Name, GenoData.value, GenoSE.error, GenoData.Id "
+ "FROM (GenoData, GenoFreeze, Strain, Geno, GenoXRef) "
+ "LEFT JOIN GenoSE ON "
+ "(GenoSE.DataId = GenoData.Id AND GenoSE.StrainId = GenoData.StrainId) "
+ "WHERE Geno.SpeciesId = %(species_id)s "
+ "AND Geno.Name = %(trait_name)s AND GenoXRef.GenoId = Geno.Id "
+ "AND GenoXRef.GenoFreezeId = GenoFreeze.Id "
+ "AND GenoFreeze.Name = %(dataset_name)s "
+ "AND GenoXRef.DataId = GenoData.Id "
+ "AND GenoData.StrainId = Strain.Id "
+ "ORDER BY Strain.Name")
+ with conn.cursor() as cursor:
+ cursor.execute(
+ query,
+ {"trait_name": trait_info["trait_name"],
+ "dataset_name": trait_info["db"]["dataset_name"],
+ "species_id": retrieve_species_id(
+ trait_info["db"]["riset"], conn)})
+ return [dict(zip(
+ ["strain_name", "value", "se_error", "id"], row))
+ for row in cursor.fetchall()]
+ return []
+
+def retrieve_publish_trait_data(trait_info: Dict, conn: Any):
+ """
+ Retrieve trait data for `Publish` traits.
+ """
+ query = (
+ "SELECT "
+ "Strain.Name, PublishData.value, PublishSE.error, NStrain.count, "
+ "PublishData.Id "
+ "FROM (PublishData, Strain, PublishXRef, PublishFreeze) "
+ "LEFT JOIN PublishSE ON "
+ "(PublishSE.DataId = PublishData.Id "
+ "AND PublishSE.StrainId = PublishData.StrainId) "
+ "LEFT JOIN NStrain ON "
+ "(NStrain.DataId = PublishData.Id "
+ "AND NStrain.StrainId = PublishData.StrainId) "
+ "WHERE PublishXRef.InbredSetId = PublishFreeze.InbredSetId "
+ "AND PublishData.Id = PublishXRef.DataId "
+ "AND PublishXRef.Id = %(trait_name)s "
+ "AND PublishFreeze.Id = %(dataset_id)s "
+ "AND PublishData.StrainId = Strain.Id "
+ "ORDER BY Strain.Name")
+ with conn.cursor() as cursor:
+ cursor.execute(
+ query,
+ {"trait_name": trait_info["trait_name"],
+ "dataset_id": trait_info["db"]["dataset_id"]})
+ return [dict(zip(
+ ["strain_name", "value", "se_error", "nstrain", "id"], row))
+ for row in cursor.fetchall()]
+ return []
+
+def retrieve_cellid_trait_data(trait_info: Dict, conn: Any):
+ """
+ Retrieve trait data for `Probe Data` types.
+ """
+ query = (
+ "SELECT "
+ "Strain.Name, ProbeData.value, ProbeSE.error, ProbeData.Id "
+ "FROM (ProbeData, ProbeFreeze, ProbeSetFreeze, ProbeXRef, Strain,"
+ " Probe, ProbeSet) "
+ "LEFT JOIN ProbeSE ON "
+ "(ProbeSE.DataId = ProbeData.Id "
+ " AND ProbeSE.StrainId = ProbeData.StrainId) "
+ "WHERE Probe.Name = %(cellid)s "
+ "AND ProbeSet.Name = %(trait_name)s "
+ "AND Probe.ProbeSetId = ProbeSet.Id "
+ "AND ProbeXRef.ProbeId = Probe.Id "
+ "AND ProbeXRef.ProbeFreezeId = ProbeFreeze.Id "
+ "AND ProbeSetFreeze.ProbeFreezeId = ProbeFreeze.Id "
+ "AND ProbeSetFreeze.Name = %(dataset_name)s "
+ "AND ProbeXRef.DataId = ProbeData.Id "
+ "AND ProbeData.StrainId = Strain.Id "
+ "ORDER BY Strain.Name")
+ with conn.cursor() as cursor:
+ cursor.execute(
+ query,
+ {"cellid": trait_info["cellid"],
+ "trait_name": trait_info["trait_name"],
+ "dataset_id": trait_info["db"]["dataset_id"]})
+ return [dict(zip(
+ ["strain_name", "value", "se_error", "id"], row))
+ for row in cursor.fetchall()]
+ return []
+
+def retrieve_probeset_trait_data(trait_info: Dict, conn: Any):
+ """
+ Retrieve trait data for `ProbeSet` traits.
+ """
+ query = (
+ "SELECT Strain.Name, ProbeSetData.value, ProbeSetSE.error, "
+ "ProbeSetData.Id "
+ "FROM (ProbeSetData, ProbeSetFreeze, Strain, ProbeSet, ProbeSetXRef) "
+ "LEFT JOIN ProbeSetSE ON "
+ "(ProbeSetSE.DataId = ProbeSetData.Id "
+ "AND ProbeSetSE.StrainId = ProbeSetData.StrainId) "
+ "WHERE ProbeSet.Name = %(trait_name)s "
+ "AND ProbeSetXRef.ProbeSetId = ProbeSet.Id "
+ "AND ProbeSetXRef.ProbeSetFreezeId = ProbeSetFreeze.Id "
+ "AND ProbeSetFreeze.Name = %(dataset_name)s "
+ "AND ProbeSetXRef.DataId = ProbeSetData.Id "
+ "AND ProbeSetData.StrainId = Strain.Id "
+ "ORDER BY Strain.Name")
+
+ with conn.cursor() as cursor:
+ cursor.execute(
+ query,
+ {"trait_name": trait_info["trait_name"],
+ "dataset_name": trait_info["db"]["dataset_name"]})
+ return [dict(zip(
+ ["strain_name", "value", "se_error", "id"], row))
+ for row in cursor.fetchall()]
+ return []
+
+def with_strainlist_data_setup(strainlist: Sequence[str]):
+ """
+ Build function that computes the trait data from provided list of strains.
+
+ PARAMETERS
+ strainlist: (list)
+ A list of strain names
+
+ RETURNS:
+ Returns a function that given some data from the database, computes the
+ strain's value, variance and ndata values, only if the strain is present
+ in the provided `strainlist` variable.
+ """
+ def setup_fn(tdata):
+ if tdata["strain_name"] in strainlist:
+ val = tdata["value"]
+ if val is not None:
+ return {
+ "strain_name": tdata["strain_name"],
+ "value": val,
+ "variance": tdata["se_error"],
+ "ndata": tdata.get("nstrain", None)
+ }
+ return None
+ return setup_fn
+
+def without_strainlist_data_setup():
+ """
+ Build function that computes the trait data.
+
+ RETURNS:
+ Returns a function that given some data from the database, computes the
+ strain's value, variance and ndata values.
+ """
+ def setup_fn(tdata):
+ val = tdata["value"]
+ if val is not None:
+ return {
+ "strain_name": tdata["strain_name"],
+ "value": val,
+ "variance": tdata["se_error"],
+ "ndata": tdata.get("nstrain", None)
+ }
+ return None
+ return setup_fn
+
+def retrieve_trait_data(trait: dict, conn: Any, strainlist: Sequence[str] = tuple()):
+ """
+ Retrieve trait data
+
+ DESCRIPTION
+ Retrieve trait data as is done in
+ https://github.com/genenetwork/genenetwork1/blob/master/web/webqtl/base/webqtlTrait.py#L258-L386
+ """
+ # I do not like this section, but it retains the flow in the old codebase
+ if trait["db"]["dataset_type"] == "Temp":
+ results = retrieve_temp_trait_data(trait, conn)
+ elif trait["db"]["dataset_type"] == "Publish":
+ results = retrieve_publish_trait_data(trait, conn)
+ elif trait["cellid"]:
+ results = retrieve_cellid_trait_data(trait, conn)
+ elif trait["db"]["dataset_type"] == "ProbeSet":
+ results = retrieve_probeset_trait_data(trait, conn)
+ else:
+ results = retrieve_geno_trait_data(trait, conn)
+
+ if results:
+ # do something with mysqlid
+ mysqlid = results[0]["id"]
+ if strainlist:
+ data = [
+ item for item in
+ map(with_strainlist_data_setup(strainlist), results)
+ if item is not None]
+ else:
+ data = [
+ item for item in
+ map(without_strainlist_data_setup(), results)
+ if item is not None]
+
+ return {
+ "mysqlid": mysqlid,
+ "data": dict(map(
+ lambda x: (
+ x["strain_name"],
+ {k:v for k, v in x.items() if x != "strain_name"}),
+ data))}
+ return {}