From 8e4e15ae62989f07d3ca8512165ba2a93c42aa94 Mon Sep 17 00:00:00 2001
From: Frederick Muriuki Muriithi
Date: Fri, 12 Apr 2024 09:17:49 +0300
Subject: Consistently check for possibly non-existent data

The "geno-dataset", "tissue", "pheno-study" and "pheno-dataset" data
"objects" might not exist for a particular uploaded bundle, so we
check in a consistent manner to ensure they are provided when needed.
---
 qc_app/db/datasets.py                    | 22 +++++++++-----------
 qc_app/templates/rqtl2/summary-info.html | 11 ++++++++++
 qc_app/upload/rqtl2.py                   | 35 +++++++++++++-------------------
 3 files changed, 35 insertions(+), 33 deletions(-)

(limited to 'qc_app')

diff --git a/qc_app/db/datasets.py b/qc_app/db/datasets.py
index 448db18..b56dcf5 100644
--- a/qc_app/db/datasets.py
+++ b/qc_app/db/datasets.py
@@ -18,12 +18,12 @@ def geno_datasets_by_species_and_population(
             {"sid": speciesid, "pid": populationid})
         return tuple(dict(row) for row in cursor.fetchall())
 
-def geno_dataset_by_id(conn: mdb.Connection, dataset_id: int) -> dict:
+def geno_dataset_by_id(conn: mdb.Connection, dataset_id) -> Optional[dict]:
     """Retrieve genotype dataset by ID"""
     with conn.cursor(cursorclass=DictCursor) as cursor:
-        cursor.execute("SELECT * FROM GenoFreeze WHERE Id=%s",
-                       (dataset_id,))
-        return dict(cursor.fetchone())
+        cursor.execute("SELECT * FROM GenoFreeze WHERE Id=%s", (dataset_id,))
+        _dataset = cursor.fetchone()
+        return dict(_dataset) if bool(_dataset) else None
 
 def probeset_studies_by_species_and_population(
         conn: mdb.Connection,
@@ -46,12 +46,12 @@ def probeset_datasets_by_study(conn: mdb.Connection,
                        (studyid,))
         return tuple(dict(row) for row in cursor.fetchall())
 
-def probeset_study_by_id(conn: mdb.Connection, studyid: int) -> dict:
+def probeset_study_by_id(conn: mdb.Connection, studyid) -> Optional[dict]:
     """Retrieve ProbeSet study by ID"""
     with conn.cursor(cursorclass=DictCursor) as cursor:
-        cursor.execute("SELECT * FROM ProbeFreeze WHERE Id=%s",
-                       (studyid,))
-        return dict(cursor.fetchone())
+        cursor.execute("SELECT * FROM ProbeFreeze WHERE Id=%s", (studyid,))
+        _study = cursor.fetchone()
+        return dict(_study) if bool(_study) else None
 
 def probeset_create_study(conn: mdb.Connection,#pylint: disable=[too-many-arguments]
                           populationid: int,
@@ -120,12 +120,10 @@ def probeset_create_dataset(conn: mdb.Connection,#pylint: disable=[too-many-argu
             dataset)
         return {**dataset, "datasetid": cursor.lastrowid}
 
-def probeset_dataset_by_id(
-        conn: mdb.Connection, datasetid: int) -> Optional[dict]:
+def probeset_dataset_by_id(conn: mdb.Connection, datasetid) -> Optional[dict]:
     """Fetch a ProbeSet dataset by its ID"""
     with conn.cursor(cursorclass=DictCursor) as cursor:
-        cursor.execute("SELECT * FROM ProbeSetFreeze WHERE Id=%s",
-                       (datasetid,))
+        cursor.execute("SELECT * FROM ProbeSetFreeze WHERE Id=%s", (datasetid,))
         result = cursor.fetchone()
         if bool(result):
             return dict(result)
diff --git a/qc_app/templates/rqtl2/summary-info.html b/qc_app/templates/rqtl2/summary-info.html
index 3882c9c..9a9b6dd 100644
--- a/qc_app/templates/rqtl2/summary-info.html
+++ b/qc_app/templates/rqtl2/summary-info.html
@@ -19,14 +19,25 @@
   <dt>Population</dt>
   <dd>{{population.InbredSetName}}</dd>
 
+  {%if geno_dataset%}
   <dt>Genotype Dataset</dt>
   <dd>{{geno_dataset.Name}} ({{geno_dataset.FullName}})</dd>
+  {%endif%}
 
+  {%if tissue%}
+  <dt>Tissue</dt>
+  <dd>{{tissue.TissueName}} ({{tissue.Name}}, {{tissue.Short_Name}})</dd>
+  {%endif%}
+
+  {%if probe_study%}
   <dt>ProbeSet Study</dt>
   <dd>{{probe_study.Name}} ({{probe_study.FullName}})</dd>
+  {%endif%}
 
+  {%if probe_dataset%}
   <dt>ProbeSet Dataset</dt>
   <dd>{{probe_dataset.Name2}} ({{probe_dataset.FullName}})</dd>
+  {%endif%}
 </dl>
 
 <form id="frm:confirm-rqtl2bundle-details"
diff --git a/qc_app/upload/rqtl2.py b/qc_app/upload/rqtl2.py
index bf392a9..8c80ef3 100644
--- a/qc_app/upload/rqtl2.py
+++ b/qc_app/upload/rqtl2.py
@@ -525,7 +525,7 @@ def create_geno_dataset(species_id: int, population_id: int):
 
         return with_errors(__thunk__,
                            partial(check_species, conn=conn),
-                           partial(check_population, conn=conn),
+                           partial(check_population, conn=conn, species_id=species_id),
                            partial(check_r_qtl2_bundle,
                                    species_id=species_id,
                                    population_id=population_id))
@@ -765,6 +765,8 @@ def create_probeset_dataset(species_id: int, population_id: int):#pylint: disabl
                 flash("Dataset full name not provided!", errorclasses)
                 return summary_page
 
+            tissue = tissue_by_id(conn, form.get("tissueid", "").strip())
+
             study = probeset_study_by_id(conn, int(form["probe-study-id"]))
             if not bool(study):
                 flash("Invalid ProbeSet study provided!", errorclasses)
@@ -806,6 +808,7 @@ def create_probeset_dataset(species_id: int, population_id: int):#pylint: disabl
                 geno_dataset=geno_dataset_by_id(
                     conn,
                     int(request.form["geno-dataset-id"])),
+                tissue=tissue,
                 study=study,
                 avgmethod=avgmethod,
                 dataset=dset)
@@ -846,6 +849,9 @@ def select_dataset_info(species_id: int, population_id: int):
             thefile = fullpath(form["rqtl2_bundle_file"])
             with ZipFile(str(thefile), "r") as zfile:
                 cdata = r_qtl2.control_data(zfile)
+
+                geno_dataset = geno_dataset_by_id(
+                    conn,form.get("geno-dataset-id", "").strip())
                 if "geno" in cdata and not bool(form.get("geno-dataset-id")):
                     return render_template(
                         "rqtl2/select-geno-dataset.html",
@@ -854,21 +860,6 @@ def select_dataset_info(species_id: int, population_id: int):
                         rqtl2_bundle_file=thefile.name,
                         datasets=geno_datasets_by_species_and_population(
                             conn, species_id, population_id))
-                geno_dataset = geno_dataset_by_id(conn, int(form["geno-dataset-id"]))
-
-                if "pheno" in cdata and not bool(form.get("probe-study-id")):
-                    tissue = tissue_by_id(conn, form.get("tissueid", "").strip())
-                    if not bool(tissue):
-                        return render_template(
-                            "rqtl2/select-tissue.html",
-                            species=species,
-                            population=population,
-                            rqtl2_bundle_file=thefile.name,
-                            geno_dataset=geno_dataset,
-                            studies=probeset_studies_by_species_and_population(
-                                conn, species_id, population_id),
-                            platforms=platforms_by_species(conn, species_id),
-                            tissues=all_tissues(conn))
 
                 tissue = tissue_by_id(conn, form.get("tissueid", "").strip())
                 if "pheno" in cdata and not bool(tissue):
@@ -883,6 +874,9 @@ def select_dataset_info(species_id: int, population_id: int):
                         platforms=platforms_by_species(conn, species_id),
                         tissues=all_tissues(conn))
 
+                probeset_study = probeset_study_by_id(
+                    conn, form.get("probe-study-id", "").strip())
+                if "pheno" in cdata and not bool(probeset_study):
                     return render_template(
                         "rqtl2/select-probeset-study-id.html",
                         species=species,
@@ -896,7 +890,9 @@ def select_dataset_info(species_id: int, population_id: int):
                 probeset_study = probeset_study_by_id(
                     conn, int(form["probe-study-id"]))
 
-                if "pheno" in cdata and not bool(form.get("probe-dataset-id")):
+                probeset_dataset = probeset_dataset_by_id(
+                    conn, form.get("probe-dataset-id", "").strip())
+                if "pheno" in cdata and not bool(probeset_dataset):
                     return render_template(
                         "rqtl2/select-probeset-dataset.html",
                         species=species,
@@ -908,16 +904,13 @@ def select_dataset_info(species_id: int, population_id: int):
                         datasets=probeset_datasets_by_study(
                             conn, int(form["probe-study-id"])),
                         avgmethods=averaging_methods(conn))
-                probeset_study = probeset_study_by_id(
-                    conn, int(form["probe-study-id"]))
-                probeset_dataset = probeset_dataset_by_id(
-                    conn, int(form["probe-dataset-id"]))
 
             return render_template("rqtl2/summary-info.html",
                                    species=species,
                                    population=population,
                                    rqtl2_bundle_file=thefile.name,
                                    geno_dataset=geno_dataset,
+                                   tissue=tissue,
                                    probe_study=probeset_study,
                                    probe_dataset=probeset_dataset)
 
-- 
cgit v1.2.3