Merge pull request #3 from unmtransinfo/kekule

Finalize badapple_classic db

badapple1_comparison/README.md

@@ -69,10 +69,8 @@ Run `bash badapple1_comparison/sh_scripts/run_generate_scaffolds.sh`. This will
 * `o_mol2scaf`: TSV file mapping compound CID to scaffold ID(s)
 
 ### (4) Initializing the DB
-1. Install postgresql with the RDKit cartridge (requires sudo):
-`apt install postgresql-14-rdkit`
-2. Run `bash badapple1_comparison/sh_scripts/create_and_load_db.sh`
-3. (Optional) Drop the activity table to save storage: 
+1. Run `bash badapple1_comparison/sh_scripts/create_and_load_db.sh`
+2. (Optional) Drop the activity table to save storage: 
 
     `psql -d badapple_classic -c "DELETE FROM activity"`
 

badapple1_comparison/sh_scripts/db/annotate_assaystats.sh

@@ -23,9 +23,6 @@ cd $REPO_DIR
 # Step 1) Generate compound activity statistics.  Populate/annotate
 # compound table with calculated assay stats.
 # (sTotal,sTested,sActive,aTested,aActive,wTested,wActive)
-psql -d $DB_NAME -c "ALTER TABLE $SCHEMA.compound ADD COLUMN nsub_total INTEGER"
-psql -d $DB_NAME -c "ALTER TABLE $SCHEMA.compound ADD COLUMN nsub_tested INTEGER"
-psql -d $DB_NAME -c "ALTER TABLE $SCHEMA.compound ADD COLUMN nsub_active INTEGER"
 psql -d $DB_NAME -c "UPDATE $SCHEMA.compound SET (nsub_total, nsub_tested, nsub_active)  = (NULL, NULL, NULL)"
 python src/annotate_db_assaystats.py \
 	--annotate_compounds \
@@ -44,21 +41,10 @@ echo "Done annotating compounds."
 # Step 2) Generate scaf activity statistics.  Populate/annotate scaffold table with calculated assay stats.
 # Scaffold table must be ALTERed to contain activity statistics.
 # (cTotal,cTested,cActive,sTotal,sTested,sActive,aTested,aActive,wTested,wActive)
-psql -d $DB_NAME -c "ALTER TABLE $SCHEMA.scaffold ADD COLUMN ncpd_total INTEGER"
-psql -d $DB_NAME -c "ALTER TABLE $SCHEMA.scaffold ADD COLUMN ncpd_tested INTEGER"
-psql -d $DB_NAME -c "ALTER TABLE $SCHEMA.scaffold ADD COLUMN ncpd_active INTEGER"
 psql -d $DB_NAME -c "UPDATE $SCHEMA.scaffold SET (ncpd_total, ncpd_tested, ncpd_active)  = (NULL, NULL, NULL)"
-psql -d $DB_NAME -c "ALTER TABLE $SCHEMA.scaffold ADD COLUMN nsub_total INTEGER"
-psql -d $DB_NAME -c "ALTER TABLE $SCHEMA.scaffold ADD COLUMN nsub_tested INTEGER"
-psql -d $DB_NAME -c "ALTER TABLE $SCHEMA.scaffold ADD COLUMN nsub_active INTEGER"
 psql -d $DB_NAME -c "UPDATE $SCHEMA.scaffold SET (nsub_total, nsub_tested, nsub_active)  = (NULL, NULL, NULL)"
-psql -d $DB_NAME -c "ALTER TABLE $SCHEMA.scaffold ADD COLUMN nass_tested INTEGER"
-psql -d $DB_NAME -c "ALTER TABLE $SCHEMA.scaffold ADD COLUMN nass_active INTEGER"
 psql -d $DB_NAME -c "UPDATE $SCHEMA.scaffold SET (nass_tested, nass_active)  = (NULL, NULL)"
-psql -d $DB_NAME -c "ALTER TABLE $SCHEMA.scaffold ADD COLUMN nsam_tested INTEGER"
-psql -d $DB_NAME -c "ALTER TABLE $SCHEMA.scaffold ADD COLUMN nsam_active INTEGER"
 psql -d $DB_NAME -c "UPDATE $SCHEMA.scaffold SET (nsam_tested, nsam_active)  = (NULL, NULL)"
-psql -d $DB_NAME -c "ALTER TABLE $SCHEMA.scaffold ADD COLUMN in_drug BOOLEAN"
 psql -d $DB_NAME -c "UPDATE $SCHEMA.scaffold SET in_drug  = NULL"
 
 

badapple1_comparison/sh_scripts/db/create_db.sh

@@ -22,7 +22,8 @@ psql -h $DB_HOST -d $DB_NAME -c "COMMENT ON DATABASE ${DB_NAME} IS '$COMMENT'"
 psql -h $DB_HOST -d $DB_NAME <<EOF
 CREATE TABLE IF NOT EXISTS $SCHEMA.scaffold (
 	id INTEGER PRIMARY KEY,
-	scafsmi VARCHAR(512) NOT NULL,
+	scafsmi VARCHAR(512) NOT NULL UNIQUE,
+	kekule_scafsmi VARCHAR(512) NOT NULL UNIQUE,
 	scaftree VARCHAR(2048),
 	ncpd_total INTEGER,
 	ncpd_tested INTEGER,

badapple1_comparison/sh_scripts/db/load_csv_files.sh

@@ -21,19 +21,20 @@ DATA_DIR=$5
 HIERS_SCRIPT="generate_scaffolds.py"
 
 # Step 1) Load scafs with scafids + scaf2scaf relationship
-# expects SCAF_TSV_FILE to have header: scaffold_id smiles  hierarchy   scaf2scaf (TSV-separated)
+# expects SCAF_TSV_FILE to have header: scaffold_id	canon_smiles	kekule_smiles	hierarchy	scaf2scaf (TSV-separated)
 # (output from ${HIERS_SCRIPT})
 SCAF_TSV_FILE="scafs.tsv"
 SCAF_TSV_PATH="$DATA_DIR/$SCAF_TSV_FILE"
 psql -h $DB_HOST -d $DB_NAME <<EOF
 CREATE TEMP TABLE temp_scaf (
     scaffold_id INTEGER PRIMARY KEY,
-    smiles VARCHAR(512) NOT NULL,
+    canon_smiles VARCHAR(512) NOT NULL UNIQUE,
+    kekule_smiles VARCHAR(512) NOT NULL UNIQUE,
     hierarchy INTEGER,
     scaf2scaf VARCHAR(2048)
 );
-\COPY temp_scaf (scaffold_id, smiles, hierarchy, scaf2scaf) FROM '$SCAF_TSV_PATH' WITH (FORMAT CSV, DELIMITER E'\t', HEADER true);
-INSERT INTO ${SCHEMA}.scaffold (id, scafsmi, scaftree) SELECT scaffold_id, smiles, scaf2scaf FROM temp_scaf;
+\COPY temp_scaf (scaffold_id, canon_smiles, kekule_smiles, hierarchy, scaf2scaf) FROM '$SCAF_TSV_PATH' WITH (FORMAT CSV, DELIMITER E'\t', HEADER true);
+INSERT INTO ${SCHEMA}.scaffold (id, scafsmi, kekule_scafsmi, scaftree) SELECT scaffold_id, canon_smiles, kekule_smiles, scaf2scaf FROM temp_scaf;
 DROP TABLE temp_scaf;
 EOF
 psql -h $DB_HOST -d $DB_NAME -c "COMMENT ON TABLE ${SCHEMA}.scaffold IS 'Scaffold definitions from HierS, see ${HIERS_SCRIPT}. Input file is ${SCAF_TSV_PATH}'"

badapple1_comparison/src/notebooks/assay_comparison.ipynb

@@ -17,7 +17,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 51,
+   "execution_count": 17,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -35,7 +35,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 52,
+   "execution_count": 18,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -55,7 +55,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 53,
+   "execution_count": 19,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -78,7 +78,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 54,
+   "execution_count": 20,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -88,7 +88,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 55,
+   "execution_count": 21,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -105,7 +105,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 56,
+   "execution_count": 22,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -123,7 +123,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 57,
+   "execution_count": 23,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -133,7 +133,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 58,
+   "execution_count": 24,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -144,7 +144,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 59,
+   "execution_count": 25,
    "metadata": {},
    "outputs": [
     {
@@ -196,7 +196,7 @@
        "0                 774         823  "
       ]
      },
-     "execution_count": 59,
+     "execution_count": 25,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -207,7 +207,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 60,
+   "execution_count": 26,
    "metadata": {},
    "outputs": [
     {
@@ -259,7 +259,7 @@
        "0                 773         823  "
       ]
      },
-     "execution_count": 60,
+     "execution_count": 26,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -277,7 +277,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 61,
+   "execution_count": 27,
    "metadata": {},
    "outputs": [],
    "source": [