methCancer-gen: a DNA methylome dataset generator for user-specified cancer type based on conditional variational autoencoder
methCancer-gen is a deep neural network-based tool for generating DNA methylome dataset based on a user-specified cancer-type. With the matrix of DNA methylation beta values and matched cancer type information as input, the methCancer-gen approximates the underlying distribution model of the input data. After model training, methylation beta value for the specified cancer type can be generated as output. methCancer-gen can address the lack of DNA methylation data and enhance efficiency in cancer related research.
- Tensorflow (>= 1.8.0)
- Python (>= 2.7)
- Python packages : numpy, pandas
Clone the repository or download source code files.
- Reference: ./preprocessing/
- Prepare or edit "cancer_type_info.csv" file having (1) cancer types to generate dataset and (2) matrix of DNA methylation beta value dataset file path for each cancer. We recommend to generate dataset in 25 cancer types whose accuracies were validated in our manuscript using TCGA data as training dataset. (They are listed in "cancer_type_info.csv" file.) The file should contain the headers and follow the format :
cancer_type,datafile
BLCA,BLCA_TP.csv
BRCA,BRCA_TP.csv
CESC,CESC_TP.csv
COAD,COAD_TP.csv
...
Each data file should contain matrix of DNA methylation beta value, where each row and column represent CpG site and sample ID, respectively :
cpg,sample_1,sample_2,...,sample_n
cg00000029,0.249737,0.464333,...,0.061501
cg00000165,0.347463,0.115849,...,0.216793
cg00000236,0.917430,0.881644,...,0.908840
...
- Use "run_preproc.sh" to perform preprocessing.
- Reference: ./methCancer_gen/
- Prepare or edit "sample_info.csv" file having (1) cancer types to generate dataset and (2) number of samples to generate for each cancer. Cancer type should be in same order listed in preprocessing step. The file should contain the headers and follow the format :
cancer_type,sample_num
BLCA,100
BRCA,100
CESC,100
COAD,100
...
- Use "run_methCancer_gen.sh" to generate DNA methylaion dataset for user-specified cancer types.
- You can get the final output "Generated_dataset_from_methCancer_gen.csv".
If you have any question or problem, please send an email to [email protected]