Skip to content

GitLab

Commit fc51f697 authored by Alessia Marcolini's avatar Alessia Marcolini
Browse files

Handle new rSNFi method on tr, ts and ts2 splits

parent 7889bf5f
Hide whitespace changes
Inline Side-by-side
Snakefile_split
View file @ fc51f697
......@@ -29,9 +29,6 @@ try:
except:
pass
output_labels_file = [f"{DATAFOLDER}/{DATASET}/{TARGET}/{SPLIT_ID}/labels_{TARGET}_tr.txt",
f"{DATAFOLDER}/{DATASET}/{TARGET}/{SPLIT_ID}/labels_{TARGET}_ts.txt"]
rule all:
input:
......@@ -39,20 +36,15 @@ rule all:
outfolder=OUTFOLDER, dataset=DATASET, target=TARGET, model=MODEL, layers=LAYERS_CONCAT, split_id=SPLIT_ID),
expand("{outfolder}/{dataset}/{target}/{model}/{split_id}/rSNF/{layers}_tr_MCC_scores.txt",
outfolder=OUTFOLDER, dataset=DATASET, target=TARGET, model=MODEL, layers=LAYERS_CONCAT, split_id=SPLIT_ID),
expand("{outfolder}/{dataset}/{target}/{model}/{split_id}/rSNFi/{layers}_tr_MCC_scores.txt",
expand("{outfolder}/{dataset}/{target}/{model}/{split_id}/rSNFi/{layers}_ts_MCC_scores.txt",
outfolder=OUTFOLDER, dataset=DATASET, target=TARGET, model=MODEL, layers=LAYERS_CONCAT, split_id=SPLIT_ID),
expand("{outfolder}/{dataset}/{target}/{model}/{split_id}/single/{layer}_tr_MCC_scores.txt",
outfolder=OUTFOLDER, dataset=DATASET, target=TARGET, model=MODEL, layer=LAYERS, split_id=SPLIT_ID)
rule create_labels_random:
input:
f"{DATAFOLDER}/{DATASET}/{old_target}/{SPLIT_ID}/labels_{old_target}_tr.txt"
output:
output_labels_file,
expand("{datafolder}/{dataset}/{target}/{split_id}/{layer}_tr.txt",
datafolder=DATAFOLDER, dataset=DATASET, target=TARGET, split_id=SPLIT_ID, layer=LAYERS),
expand("{datafolder}/{dataset}/{target}/{split_id}/{layer}_ts.txt",
datafolder=DATAFOLDER, dataset=DATASET, target=TARGET, split_id=SPLIT_ID, layer=LAYERS)
f"{DATAFOLDER}/{DATASET}/{TARGET}/{SPLIT_ID}/labels_{TARGET}_tr.txt",
f"{DATAFOLDER}/{DATASET}/{TARGET}/{SPLIT_ID}/labels_{TARGET}_ts.txt",
run:
if RANDOM:
path = os.path.abspath(os.path.join(f'{DATAFOLDER}/{DATASET}/{old_target}/{SPLIT_ID}'))
......@@ -62,33 +54,46 @@ rule create_labels_random:
files_to_link = [f for f in os.listdir(path) if not f.startswith('labels')]
for f in files_to_link:
subprocess.call(f'ln -s {path}/{f} {path_random}/{f}', shell=True)
if not os.path.exists(f'{path_random}/{f}'):
subprocess.call(f'ln -s {path}/{f} {path_random}/{f}', shell=True)
subprocess.call(f'ln -s {path}/labels_{old_target}_ts2.txt {path_random}/labels_{TARGET}_ts2.txt', shell=True)
subprocess.call(f'ln -s {path}/labels_{old_target}_ts.txt {path_random}/labels_{TARGET}_ts.txt', shell=True)
labels_file_tr = f'{path}/labels_{old_target}_tr.txt'
shuffled_labels_file_tr = f'{path_random}/labels_{TARGET}_tr.txt'
labels_file_ts = f'{path}/labels_{old_target}_ts.txt'
shuffled_labels_file_ts = f'{path_random}/labels_{TARGET}_ts.txt'
np.random.seed(0)
with open(labels_file_tr, 'r') as f_old:
labels = np.array(f_old.readlines())
np.random.shuffle(labels)
labels_file = f'{path}/labels_{old_target}_tr.txt'
shuffled_labels_file = f'{path_random}/labels_{TARGET}_tr.txt'
with open(shuffled_labels_file_tr, 'w') as f_new:
f_new.write(''.join(labels))
np.random.seed(0)
with open(labels_file, 'r') as f_old:
with open(labels_file_ts, 'r') as f_old:
labels = np.array(f_old.readlines())
np.random.shuffle(labels)
with open(shuffled_labels_file, 'w') as f_new:
with open(shuffled_labels_file_ts, 'w') as f_new:
f_new.write(''.join(labels))
rule concat_layers:
input:
rules.create_labels_random.output,
expand("{datafolder}/{dataset}/{target}/{split_id}/{layer}_tr.txt",
datafolder=DATAFOLDER, dataset=DATASET, target=TARGET, split_id=SPLIT_ID, layer=LAYERS),
expand("{datafolder}/{dataset}/{target}/{split_id}/{layer}_ts.txt",
datafolder=DATAFOLDER, dataset=DATASET, target=TARGET, split_id=SPLIT_ID, layer=LAYERS),
expand("{datafolder}/{dataset}/{target}/{split_id}/{layer}_ts2.txt",
datafolder=DATAFOLDER, dataset=DATASET, target=TARGET, split_id=SPLIT_ID, layer=LAYERS)
output:
expand("{datafolder}/{dataset}/{target}/{split_id}/{layer}_tr.txt",
datafolder=DATAFOLDER, dataset=DATASET, target=TARGET, split_id=SPLIT_ID, layer=LAYERS_CONCAT),
expand("{datafolder}/{dataset}/{target}/{split_id}/{layer}_ts.txt",
datafolder=DATAFOLDER, dataset=DATASET, target=TARGET, split_id=SPLIT_ID, layer=LAYERS_CONCAT),
expand("{datafolder}/{dataset}/{target}/{split_id}/{layer}_ts2.txt",
datafolder=DATAFOLDER, dataset=DATASET, target=TARGET, split_id=SPLIT_ID, layer=LAYERS_CONCAT)
shell:
f"python preprocessing/concat_layers.py --datafolder {DATAFOLDER} --dataset {DATASET} --target {TARGET} --layers {LAYERS_SPACED} --split_id {SPLIT_ID}"
......@@ -124,7 +129,7 @@ rule snf:
threads: 8
output:
expand("{outfolder}/{dataset}/{target}/{model}/{split_id}/rSNF/INF_{layers}_tr.txt",
outfolder=OUTFOLDER, dataset=DATASET, target=TARGET, model=MODEL, layers=LAYERS_CONCAT, split_id=SPLIT_ID)
outfolder=OUTFOLDER, dataset=DATASET, target=TARGET, model=MODEL, layers=LAYERS_CONCAT, split_id=SPLIT_ID)
run:
all_input = [i[1] for i in input.allitems()]
inputfiles = " ".join(all_input[:-1])
......@@ -176,12 +181,12 @@ rule myintersect:
rule extract:
input:
expand("{datafolder}/{dataset}/{target}/{split_id}/{layers}_tr.txt",
expand("{datafolder}/{dataset}/{target}/{split_id}/{layers}_ts.txt",
datafolder=DATAFOLDER, dataset=DATASET, target=TARGET, layers=LAYERS_CONCAT, split_id=SPLIT_ID),
expand("{outfolder}/{dataset}/{target}/{model}/{split_id}/rSNFi/{layers}_intersect_tr.txt",
outfolder=OUTFOLDER, dataset=DATASET, target=TARGET, model=MODEL, layers=LAYERS_CONCAT, split_id=SPLIT_ID)
output:
expand("{outfolder}/{dataset}/{target}/{model}/{split_id}/rSNFi/{layers}_tr.txt",
expand("{outfolder}/{dataset}/{target}/{model}/{split_id}/rSNFi/{layers}_ts.txt",
outfolder=OUTFOLDER, dataset=DATASET, target=TARGET, model=MODEL, layers=LAYERS_CONCAT, split_id=SPLIT_ID)
shell:
"python extract_topfeats_onecol.py {input} {output}"
......@@ -189,26 +194,26 @@ rule extract:
rule ml_rsnfi_tr:
input:
expand("{outfolder}/{dataset}/{target}/{model}/{split_id}/rSNFi/{layers}_tr.txt",
expand("{outfolder}/{dataset}/{target}/{model}/{split_id}/rSNFi/{layers}_ts.txt",
outfolder=OUTFOLDER, dataset=DATASET, target=TARGET, model=MODEL, split_id=SPLIT_ID, layers=LAYERS_CONCAT),
expand("{datafolder}/{dataset}/{target}/{split_id}/labels_{target}_tr.txt",
expand("{datafolder}/{dataset}/{target}/{split_id}/labels_{target}_ts.txt",
datafolder=DATAFOLDER, dataset=DATASET, target=TARGET, split_id=SPLIT_ID, layers=LAYERS_CONCAT),
output:
"{outfolder}/{dataset}/{target}/{model}/{split_id}/rSNFi/{layers}_tr_{model}_KBest.log"
"{outfolder}/{dataset}/{target}/{model}/{split_id}/rSNFi/{layers}_ts_{model}_KBest.log"
shell:
"python sklearn_training.py {input} {wildcards.outfolder}/{wildcards.dataset}/{wildcards.target}/{wildcards.model}/{wildcards.split_id}/rSNFi --model {wildcards.model} --ranking KBest"
rule ml_rsnfi_val:
input:
expand("{outfolder}/{dataset}/{target}/{model}/{split_id}/rSNFi/{layers}_tr_{model}_KBest.log",
expand("{outfolder}/{dataset}/{target}/{model}/{split_id}/rSNFi/{layers}_ts_{model}_KBest.log",
outfolder=OUTFOLDER, dataset=DATASET, target=TARGET, model=MODEL, split_id=SPLIT_ID, layers=LAYERS_CONCAT),
expand("{datafolder}/{dataset}/{target}/{split_id}/{layers}_ts.txt",
expand("{datafolder}/{dataset}/{target}/{split_id}/{layers}_ts2.txt",
datafolder=DATAFOLDER, dataset=DATASET, target=TARGET, split_id=SPLIT_ID, layers=LAYERS_CONCAT),
expand("{datafolder}/{dataset}/{target}/{split_id}/labels_{target}_ts.txt",
expand("{datafolder}/{dataset}/{target}/{split_id}/labels_{target}_ts2.txt",
datafolder=DATAFOLDER, dataset=DATASET, target=TARGET, split_id=SPLIT_ID, layers=LAYERS_CONCAT)
output:
"{outfolder}/{dataset}/{target}/{model}/{split_id}/rSNFi/{layers}_tr_MCC_scores.txt"
"{outfolder}/{dataset}/{target}/{model}/{split_id}/rSNFi/{layers}_ts_MCC_scores.txt"
shell:
"python sklearn_validation.py {input[0]} {input[1]} {wildcards.outfolder}/{wildcards.dataset}/{wildcards.target}/{wildcards.model}/{wildcards.split_id}/rSNFi --tslab {input[2]}"
......
postprocessing/input_output.py 0 → 100644
View file @ fc51f697
import numpy as np
import pandas as pd
def load_data(filename):
df = pd.read_csv(filename, sep='\t', header=0, index_col=0)
var_names = df.columns.tolist()
sample_names = df.index.tolist()
data = df.values.astype(dtype=np.float)
return sample_names, var_names, data
def save_split(x, y, sample_names, var_names, basename):
"""
x, y: output of train_test_split
sample_names var_names: lists with samples and feature names (will be the DataFrame row and column names)
"""
x_df = pd.DataFrame(x, index=sample_names, columns=var_names)
x_df.to_csv(f"{basename}.txt", sep='\t', index=True, header=True, index_label="sampleID")
y_df = pd.DataFrame(y, index=sample_names, columns=['label'])
y_df.to_csv(f"{basename}.lab", sep='\t', index=True, header=True, index_label="sampleID")
preprocessing/concat_layers.py
View file @ fc51f697
......@@ -49,16 +49,20 @@ for k in range(2, len(LAYERS) + 1):
for comb in combinations(LAYERS, k):
single_dfs_tr = []
single_dfs_ts = []
single_dfs_ts2 = []
for layer in comb:
single_dfs_tr.append(pd.read_csv(f'{PATH}/{layer}_tr.txt', sep='\t'))
single_dfs_ts.append(pd.read_csv(f'{PATH}/{layer}_ts.txt', sep='\t'))
single_dfs_ts2.append(pd.read_csv(f'{PATH}/{layer}_ts2.txt', sep='\t'))
merged_tr = reduce(lambda x, y: pd.merge(x, y, on='Sample'), single_dfs_tr)
merged_ts = reduce(lambda x, y: pd.merge(x, y, on='Sample'), single_dfs_ts)
merged_ts2 = reduce(lambda x, y: pd.merge(x, y, on='Sample'), single_dfs_ts2)
layers_concat = '_'.join(comb)
merged_tr.to_csv(f'{PATH}/{layers_concat}_tr.txt', sep='\t', index=False)
merged_ts.to_csv(f'{PATH}/{layers_concat}_ts.txt', sep='\t', index=False)
merged_ts2.to_csv(f'{PATH}/{layers_concat}_ts2.txt', sep='\t', index=False)
# %%
resplitter.py 0 → 100644
View file @ fc51f697
import argparse
import os
import subprocess
from functools import reduce
import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
parser = argparse.ArgumentParser()
parser.add_argument('--datafolder', type=str, help='Data folder')
parser.add_argument('--dataset', type=str, help='Dataset name')
parser.add_argument('--target', type=str, help='Clinical endpoint')
parser.add_argument('--n_splits_end', type=int, help='')
parser.add_argument('--n_splits_start', type=int, help='')
parser.add_argument('--split_offset', type=int, default=50, help='')
args = parser.parse_args()
#%%
DATAFOLDER = args.datafolder
DATASET = args.dataset
TARGET = args.target
N_SPLITS_START = args.n_splits_start
N_SPLITS_END = args.n_splits_end
SPLIT_OFFSET = args.split_offset
assert (
SPLIT_OFFSET > N_SPLITS_END - N_SPLITS_START
), 'New splits set must not overlap with old splits set'
########### calculate new tr ratio
tmp_y_tr = pd.read_csv(
f'{DATAFOLDER}/{DATASET}/{TARGET}/{N_SPLITS_START}/labels_{TARGET}_tr.txt',
sep='\t',
header=None,
)
tmp_y_ts = pd.read_csv(
f'{DATAFOLDER}/{DATASET}/{TARGET}/{N_SPLITS_START}/labels_{TARGET}_ts.txt',
sep='\t',
header=None,
)
total_dataset_len = tmp_y_tr.shape[0] + tmp_y_ts.shape[0]
ts2_dataset_len = round(0.2 * total_dataset_len)
ts2_ratio = round(ts2_dataset_len / tmp_y_tr.shape[0], 2)
###########
for split_old in range(N_SPLITS_START, N_SPLITS_END):
split_new = split_old + SPLIT_OFFSET
path_old = f'{DATAFOLDER}/{DATASET}/{TARGET}/{split_old}'
path_new = f'{DATAFOLDER}/{DATASET}/{TARGET}/{split_new}'
subprocess.call(f'rsync -rav {path_old}/ {path_new}', shell=True)
labelsfile = f"labels_{TARGET}_tr.txt"
print('resplitter')
y_orig = pd.read_csv(f'{path_new}/{labelsfile}', header=None)
for f in os.listdir(path_new):
if f.endswith("_tr.txt"):
if not f.startswith("labels") and not f.startswith("clin"):
print(f)
df = pd.read_csv(f'{path_new}/{f}', sep="\t", index_col=0)
df_tr_new, df_ts2, y_tr_new, y_ts2 = train_test_split(
df, y_orig, test_size=ts2_ratio, random_state=0, stratify=y_orig
)
df_tr_new.to_csv(f'{path_new}/{f}', sep="\t")
df_ts2.to_csv(
f'{path_new}/{f.replace("_tr.txt", "_ts2.txt")}', sep="\t"
)
y_tr_new.to_csv(
f'{path_new}/labels_{TARGET}_tr.txt',
sep="\t",
index=False,
header=False,
)
y_ts2.to_csv(
f'{path_new}/labels_{TARGET}_ts2.txt',
sep="\t",
index=False,
header=False,
)
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment