From 434a9647181578976a8cf53bd195046846380f34 Mon Sep 17 00:00:00 2001
From: strasserle <strasser.lena@gmx.de>
Date: Sat, 11 Jan 2025 00:05:58 +0100
Subject: [PATCH] query /alternativeSplicing/getPsiValue/ now also by dataset

---
 app/controllers/alternativeSplicing.py | 18 ++++++++++++++++--
 swagger.yml                            | 19 +++++++++++++++++++
 2 files changed, 35 insertions(+), 2 deletions(-)

diff --git a/app/controllers/alternativeSplicing.py b/app/controllers/alternativeSplicing.py
index c824c9e..a033f45 100644
--- a/app/controllers/alternativeSplicing.py
+++ b/app/controllers/alternativeSplicing.py
@@ -1,7 +1,9 @@
 from flask import abort
+from sqlalchemy import or_
+from app.controllers.dataset import _dataset_query
 import app.models as models
 from flask import Response
-from app.config import db
+from app.config import db, LATEST
 
 def get_transcript_events(enst_number):
     """
@@ -93,16 +95,26 @@ def get_exons_for_position(start_pos: int, end_pos: int):
         abort(404, "No data found that satisfies the given filters")
 
 
-def get_psi_values(transcript_ID: str = None, enst_number: str =None, psivec_ID: int = None, alternative_splicing_event_transcripts_ID: str = None, sample_ID: str = None, limit=100):
+def get_psi_values(dataset_ID: str = None, disease_name: str = None, data_origin: str = None, transcript_ID: str = None, enst_number: str =None, psivec_ID: int = None, alternative_splicing_event_transcripts_ID: str = None, sample_ID: str = None, limit=100, sponge_db_version: int = LATEST):
     """
     This function response for the request: /alternativeSplicing/getPsiValue/
     with the possibility to filter by psivec_ID, alternative
     splicing event transcripts ID and sample ID
+    :param disease_name: name of the disease
+    :param dataset_ID: ID of the dataset
     :param psivec_ID: ID of the psivec
     :param alternative_splicing_event_transcripts_ID: ID of the alternative splicing event transcripts
     :param sample_ID: ID of the sample
     :return: psi value for the given parameters, ordered by psi value
     """
+
+    # Get the dataset
+    if dataset_ID or disease_name or data_origin:
+        data = _dataset_query(sponge_db_version=sponge_db_version, dataset_ID=dataset_ID, disease_name=disease_name, data_origin=data_origin)
+        data = [d.disease_name for d in data]
+    else: 
+        data = None
+
     # Build the transcript query
     transcript_query = db.select(models.Transcript.transcript_ID)
     if transcript_ID:
@@ -127,6 +139,8 @@ def get_psi_values(transcript_ID: str = None, enst_number: str =None, psivec_ID:
         psi_query = psi_query.where(models.PsiVec.psivec_ID == psivec_ID)
     if sample_ID:
         psi_query = psi_query.where(models.PsiVec.sample_ID == sample_ID)
+    if data:
+        psi_query = psi_query.where(or_(*[models.PsiVec.sample_ID.like(d.replace(" ", "_") + "%") for d in data]))
 
     # Apply limit and sort results
     psi_query = psi_query.order_by(models.PsiVec.psi_value.desc()).limit(limit)
diff --git a/swagger.yml b/swagger.yml
index ebd56c5..5094fc9 100644
--- a/swagger.yml
+++ b/swagger.yml
@@ -3953,6 +3953,24 @@ paths:
       summary: Retrieve all psi values for a specific event type of a transcript of interest.
       description: Retrieve all psi values for a specific event type of a transcript of interest.
       parameters:
+      - name: dataset_ID
+        in: query
+        description: Internal database ID of the cancer type/dataset.
+        required: false
+        schema:
+          type: integer
+      - name: disease_name
+        in: query
+        description: Name of the specific cancer type/dataset. Fuzzy search is available (e.g. "kidney clear cell carcinoma" or just "kidney").
+        required: false
+        schema:
+          type: string
+      - name: data_origin
+        in: query
+        description: Origin of the data (e.g. TCGA, GTEx, ...)
+        required: false
+        schema:
+          type: string
       - name: transcript_ID
         in: query
         description: Internal database ID of the transcript of interest.
@@ -3989,6 +4007,7 @@ paths:
         schema:
           type: integer
         required: false
+      - $ref: '#/components/parameters/VersionParam'
       responses:
         "200":
           description: Retrieve all psi values for a specific event type of a transcript of interest.