manage custom splits

postprocessing/borda_global_juxt_rSNF.py

@@ -39,7 +39,8 @@ parser.add_argument('--dataset', type=str, help='Dataset name')
 parser.add_argument('--target', type=str, help='Clinical endpoint')
 parser.add_argument('--model', type=str, help='Classifiers implemented, randomForest or LSVM')
 parser.add_argument('--layers', type=str, nargs='+', help='')
-parser.add_argument('--n_splits', type=int, help='')
+parser.add_argument('--n_splits_end', type=int, help='')
+parser.add_argument('--n_splits_start', type=int, help='')
 parser.add_argument('--mode', type=str, help='rSNF, rSNFi, single')
 
 args = parser.parse_args()
@@ -51,7 +52,8 @@ OUTFOLDER = args.outfolder
 TARGET = args.target
 MODEL = args.model
 LAYERS = args.layers
-N_SPLITS = args.n_splits
+N_SPLITS_START = args.n_splits_start
+N_SPLITS_END = args.n_splits_end
 MODE = args.mode
 assert MODE in ['juxt', 'rSNF', 'single']
 
@@ -66,7 +68,7 @@ for k in range(2, N_LAYERS + 1):
         _, var_names, _ = load_data(os.path.join(DATAFOLDER, DATASET, TARGET, f'0/{layers_concat}_tr.txt') )
         rankings = [] 
 
-        for i in range(N_SPLITS):
+        for i in range(N_SPLITS_END-N_SPLITS_START):
 
             if MODE == 'rSNF':  
                 file_ranking = os.path.join(OUTFOLDER, DATASET, TARGET, MODEL, f'{i}/{MODE}/{layers_concat}_tr_{MODEL}_rankList_ranking.csv.gz')    
@@ -83,4 +85,4 @@ for k in range(2, N_LAYERS + 1):
         for i, pos in zip(BORDA_ID, BORDA_POS):
             borda_df = borda_df.append({'FEATURE_ID': i, 'FEATURE_NAME': var_names[i], 'MEAN_POS': pos+1}, ignore_index=True)
 
-        borda_df.to_csv(f"{OUTFOLDER}/{DATASET}/{TARGET}/{MODEL}/Borda_allSplits_{MODE}_{layers_concat}.txt", sep='\t', index=False, float_format="%.3f")
+        borda_df.to_csv(f"{OUTFOLDER}/{DATASET}/{TARGET}/{MODEL}/Borda_splits_{N_SPLITS_START}-{N_SPLITS_END}_{MODE}_{layers_concat}.txt", sep='\t', index=False, float_format="%.3f")

postprocessing/borda_global_rSNFi.py

@@ -29,7 +29,8 @@ parser.add_argument('--dataset', type=str, help='Dataset name')
 parser.add_argument('--target', type=str, help='Clinical endpoint')
 parser.add_argument('--model', type=str, help='Classifiers implemented, randomForest or LSVM')
 parser.add_argument('--layers', type=str, nargs='+', help='')
-parser.add_argument('--n_splits', type=int, help='')
+parser.add_argument('--n_splits_end', type=int, help='')
+parser.add_argument('--n_splits_start', type=int, help='')
 parser.add_argument('--mode', type=str, help='rSNFi custom Borda')
 
 args = parser.parse_args()
@@ -41,7 +42,8 @@ OUTFOLDER = args.outfolder
 TARGET = args.target
 MODEL = args.model
 LAYERS = args.layers
-N_SPLITS = args.n_splits
+N_SPLITS_START = args.n_splits_start
+N_SPLITS_END = args.n_splits_end
 MODE = args.mode
 assert MODE == 'rSNFi'
 
@@ -59,7 +61,7 @@ for k in range(2, N_LAYERS + 1):
 
         all_feats=[]
 
-        for i in range(N_SPLITS):
+        for i in range(N_SPLITS_END-N_SPLITS_START):
             file_featureList = os.path.join(OUTFOLDER, DATASET, TARGET, MODEL, f'{i}/{MODE}/{layers_concat}_tr_{MODEL}_KBest_featurelist.txt') 
             feats = pd.read_csv(file_featureList, sep='\t')
             all_feats.extend(list(feats.FEATURE_NAME))
@@ -68,13 +70,13 @@ for k in range(2, N_LAYERS + 1):
 
         positions = dict()
         means = dict()
-        x=((len(all_feats)-1)*np.ones((1,N_SPLITS*CV_K*CV_N)))
+        x=((len(all_feats)-1)*np.ones((1,(N_SPLITS_END-N_SPLITS_START)*CV_K*CV_N)))
 
         for i in all_feats:
             positions[i]=x.tolist()[0]
             means[i]=0.0
 
-        for i in range(N_SPLITS):
+        for i in range(N_SPLITS_END-N_SPLITS_START):
 
             file_featureList = os.path.join(OUTFOLDER, DATASET, TARGET, MODEL, f'{i}/{MODE}/{layers_concat}_tr_{MODEL}_KBest_featurelist.txt') 
             feats = pd.read_csv(file_featureList, sep='\t')
@@ -96,4 +98,4 @@ for k in range(2, N_LAYERS + 1):
         sorted_means = sorted(means.items(), key=operator.itemgetter(1))
         
         borda_df = pd.DataFrame(sorted_means, columns=['FEATURE_NAME', 'MEAN_POS'])
-        borda_df.to_csv(f"{OUTFOLDER}/{DATASET}/{TARGET}/{MODEL}/Borda_allSplits_{MODE}_{layers_concat}.txt", sep='\t', index=False, float_format="%.3f")
\ No newline at end of file
+        borda_df.to_csv(f"{OUTFOLDER}/{DATASET}/{TARGET}/{MODEL}/Borda_splits_{N_SPLITS_START}-{N_SPLITS_END}_{MODE}_{layers_concat}.txt", sep='\t', index=False, float_format="%.3f")
\ No newline at end of file

postprocessing/compute_all_metrics.py

@@ -35,7 +35,8 @@ parser.add_argument(
     '--model', type=str, help='classifiers implemented, randomForest or LSVM'
 )
 parser.add_argument('--layers', type=str, nargs='+', help='')
-parser.add_argument('--n_splits', type=int, help='')
+parser.add_argument('--n_splits_end', type=int, help='')
+parser.add_argument('--n_splits_start', type=int, help='')
 parser.add_argument('--mode', type=str, help='juxt, rSNF, rSNFi, single')
 
 args = parser.parse_args()
@@ -46,7 +47,8 @@ DATASET = args.dataset
 TARGET = args.target
 MODEL = args.model
 LAYERS = args.layers
-N_SPLITS = args.n_splits
+N_SPLITS_START = args.n_splits_start
+N_SPLITS_END = args.n_splits_end
 MODE = args.mode
 
 assert MODE in ['juxt', 'rSNF', 'rSNFi', 'single']
@@ -98,7 +100,7 @@ for k in range_combinations:
         all_test_mccs = []
         best_feat_steps = []
 
-        for split_id in range(N_SPLITS):
+        for split_id in range(N_SPLITS_END-N_SPLITS_START):
 
             PATH = f'{OUTFOLDER}/{DATASET}/{TARGET}/{MODEL}/{split_id}'
 
@@ -255,7 +257,7 @@ for k in range_combinations:
         df_results = df_results.append(row, ignore_index=True)
 
 df_results.to_csv(
-    f'{OUTFOLDER}/{DATASET}/{TARGET}/{MODEL}/metrics_allSplits_{MODE}.txt',
+    f'{OUTFOLDER}/{DATASET}/{TARGET}/{MODEL}/metrics_splits_{N_SPLITS_START}-{N_SPLITS_END}_{MODE}.txt',
     sep='\t',
     index=False,
 )

runner.sh

@@ -5,33 +5,34 @@ THREADS=12
 
 OUTFOLDER=results
 DATAFOLDER=data
-DATASET=tcga_aml
-LAYER1=gene
-LAYER2=meth
-LAYER3=mirna
-TARGET=OS
+DATASET=tcga_breast
+LAYER1=cnv
+LAYER2=prot
+LAYER3=prot
+TARGET=ER
 MODEL=LSVM
-N_SPLITS=10
+N_SPLITS_START=0
+N_SPLITS_END=10
 
 # go!
 
-for (( i=0; i<$N_SPLITS; i++ ))
+for (( i=$N_SPLITS_START; i<$N_SPLITS_END; i++ ))
 do
-    snakemake -s Snakefile_split --cores $THREADS --config datafolder=$DATAFOLDER outfolder=$OUTFOLDER dataset=$DATASET target=$TARGET layer1=$LAYER1 layer2=$LAYER2 layer3=$LAYER3 model=$MODEL split_id=$i -p 
+    snakemake -s Snakefile_split --cores $THREADS --config datafolder=$DATAFOLDER outfolder=$OUTFOLDER dataset=$DATASET target=$TARGET layer1=$LAYER1 layer2=$LAYER2  model=$MODEL split_id=$i -p 
 done
 
 for MODE in juxt rSNF rSNFi single
 do
-    python postprocessing/compute_all_metrics.py --outfolder $OUTFOLDER --dataset $DATASET --target $TARGET --layers $LAYER1 $LAYER2 $LAYER3 --model $MODEL --n_splits $N_SPLITS --mode $MODE
+    python postprocessing/compute_all_metrics.py --outfolder $OUTFOLDER --dataset $DATASET --target $TARGET --layers $LAYER1 $LAYER2 --model $MODEL --n_splits $N_SPLITS_END $N_SPLITS_START --mode $MODE
 done
 
 
 for MODE in juxt rSNF 
 do
-    python postprocessing/borda_global_juxt_rSNF.py --datafolder $DATAFOLDER --outfolder $OUTFOLDER --dataset $DATASET --target $TARGET --layers $LAYER1 $LAYER2 $LAYER3 --model $MODEL --n_splits $N_SPLITS --mode $MODE
+    python postprocessing/borda_global_juxt_rSNF.py --datafolder $DATAFOLDER --outfolder $OUTFOLDER --dataset $DATASET --target $TARGET --layers $LAYER1 $LAYER2 --model $MODEL --n_splits $N_SPLITS_END $N_SPLITS_START --mode $MODE
 done
 
-python postprocessing/borda_global_rSNFi.py --datafolder $DATAFOLDER --outfolder $OUTFOLDER --dataset $DATASET --target $TARGET --layers $LAYER1 $LAYER2 $LAYER3  --model $MODEL --n_splits $N_SPLITS --mode rSNFi
+python postprocessing/borda_global_rSNFi.py --datafolder $DATAFOLDER --outfolder $OUTFOLDER --dataset $DATASET --target $TARGET --layers $LAYER1 $LAYER2 --model $MODEL --n_splits $N_SPLITS_END $N_SPLITS_START --mode rSNFi