ConsDeepSignaling

In this study, we proposed a not fully connected deep learning model, ConsDeepSignaling, for drug effect prediction.

1.Data Preprocess

This study intergrates following datasets

GDSC data sets are included in folder /GDSC, and other data sets are included in /datainfo/init_data and /datainfo/mid_data.

In the main function of parse_file.py, we can adjust k to choose number of folds and choose place_num to get certain split of data set.

Finally, those datasets files will be parsed into numpy files to train our ConsDeepSignaling model.

python3 parse_file.py
python3 load_data.py

Run the code

python3 main.py

Analyze the experiment results and plot figures

python3 analysis.py

Name		Name	Last commit message	Last commit date
Latest commit History 20 Commits
GDSC		GDSC
Supplementary-Materials		Supplementary-Materials
datainfo		datainfo
flabel		flabel
.gitignore		.gitignore
README.md		README.md
analysis-5fold.py		analysis-5fold.py
analysis.py		analysis.py
analysis_dnn_simple.py		analysis_dnn_simple.py
keras_dnn_simple.py		keras_dnn_simple.py
keras_rand_nn.py		keras_rand_nn.py
load_data.py		load_data.py
main.py		main.py
main_dnn_simple.py		main_dnn_simple.py
output.py		output.py
parse_file.py		parse_file.py
reform.py		reform.py
statistic.py		statistic.py
test_regression.py		test_regression.py