From 0cf1f6287f59eb9adc69b9c50b6fee241b843361 Mon Sep 17 00:00:00 2001
From: James Kent <jamesdkent21@gmail.com>
Date: Mon, 4 Nov 2024 16:43:46 -0600
Subject: [PATCH] wip: add feature tables

---
 store/neurostore/models/data.py    | 51 +++++++++++++++++++++++++++++-
 store/neurostore/tests/conftest.py | 27 ++++++++++++++++
 2 files changed, 77 insertions(+), 1 deletion(-)

diff --git a/store/neurostore/models/data.py b/store/neurostore/models/data.py
index a3866d21..41d37118 100644
--- a/store/neurostore/models/data.py
+++ b/store/neurostore/models/data.py
@@ -278,7 +278,7 @@ class Study(BaseMixin, db.Model):
     public = db.Column(db.Boolean, default=True)
     level = db.Column(db.String)
     metadata_ = db.Column(JSONB)
-    source = db.Column(db.String, index=True)
+    source = db.Column(db.String, index=True)base_study_id = db.Column(db.Text, db.ForeignKey("base_studies.id"), index=True)
     source_id = db.Column(db.String, index=True)
     source_updated_at = db.Column(db.DateTime(timezone=True))
     base_study_id = db.Column(db.Text, db.ForeignKey("base_studies.id"), index=True)
@@ -538,6 +538,55 @@ class PointValue(BaseMixin, db.Model):
     user = relationship("User", backref=backref("point_values", passive_deletes=True))
 
 
+class Pipeline(BaseMixin, db.Model):
+    __tablename__ = "pipelines"
+
+    name = db.Column(db.String)
+    description = db.Column(db.String)
+    version = db.Column(db.String)
+    study_depenedent = db.Column(db.Boolean, default=False)
+    ace_compatible = db.Column(db.Boolean, default=False)
+    pubget_compatible = db.Column(db.Boolean, default=False)
+    derived_from = db.Column(db.Text)
+
+
+class PipelineConfig(BaseMixin, db.Model):
+    __tablename__ = "pipeline_configs"
+
+    pipeline_id = db.Column(
+        db.Text, db.ForeignKey("pipelines.id", ondelete="CASCADE"), index=True
+    )
+    config = db.Column(JSONB)
+    config_hash = db.Column(db.String, index=True)
+    pipeline = relationship("Pipeline", backref=backref("configs", passive_deletes=True))
+
+
+class PipelineRun(BaseMixin, db.Model):
+    __tablename__ = "pipeline_runs"
+
+    pipeline_id = db.Column(
+        db.Text, db.ForeignKey("pipelines.id", ondelete="CASCADE"), index=True
+    )
+    config_id = db.Column(
+        db.Text, db.ForeignKey("pipeline_configs.id", ondelete="CASCADE"), index=True
+    )
+    config = relationship("PipelineConfig", backref=backref("runs", passive_deletes=True))
+    run_index = db.Column(db.Integer())
+
+
+class PipelineRunResult(BaseMixin, db.Model):
+    __tablename__ = "pipeline_run_results"
+
+    run_id = db.Column(
+        db.Text, db.ForeignKey("pipeline_runs.id", ondelete="CASCADE"), index=True
+    )
+    base_study_id = db.Column(db.Text, db.ForeignKey("base_studies.id"), index=True)
+    feature_index = db.Column(db.Integer)  # the same categories of information can be extracted multiple times from a single paper (e.g., multiple demographic groups, multiple software packages, etc)
+    feature_group = db.Column(db.String)  # task, disease, software, age
+    feature = db.Column(db.String)  # stroop task, schizophrenia, fsl
+    value = db.Column(db.Float)  # 0.67, 0.3, 0.5 (some measure of confidence for the result)
+    run = relationship("PipelineRun", backref=backref("results", passive_deletes=True))
+
 # from . import event_listeners  # noqa E402
 
 # del event_listeners
diff --git a/store/neurostore/tests/conftest.py b/store/neurostore/tests/conftest.py
index 099ff129..7b896383 100644
--- a/store/neurostore/tests/conftest.py
+++ b/store/neurostore/tests/conftest.py
@@ -1,4 +1,6 @@
 import pytest
+import random
+import json
 from os import environ
 from neurostore.models.data import Analysis, Condition
 from sqlalchemy.orm import scoped_session, sessionmaker
@@ -586,3 +588,28 @@ def simple_neurosynth_annotation(session, ingest_neurosynth):
     session.commit()
 
     return smol_annot
+
+
+@pytest.fixture(scope="function")
+def create_demographic_features(session, ingest_neurosynth, tmp_path):
+    output_dir = tmp_path / "output" / "demographics" / "v1.0.0"
+    output_dir.mkdir(exist_ok=True, parents=True)
+    studies = Study.query.all()
+    diseases = ["schizophrenia", "bipolar disorder", "depression", "healthy"]
+    studies_data = [
+        [
+            {
+                "age": random.randint(18, 100),
+                "group": group
+            } for group in random.sample(diseases, k=random.randint(1, 2))
+        ] for study in studies
+    ]
+
+    for study, study_data in zip(studies, studies_data):
+        study_dir = output_dir / study.id
+        with open(study_dir / "results.json", "w") as f:
+            for entry in study_data:
+                json.dump(entry, f)
+                f.write('\n')
+
+    return output_dir