HCC_Detection_Paper

Code for the paper "Machine learning-enabled early detection of hepatocellular carcinoma utilizing cell-free DNA mutation and fragmentation multiplicity: a prospective study"

First download and unzip the PREDICT_Dataset.zip file

Code and dataset deposited in this repository main contained the following parts:

Key discoveries during the research:

1. The mutation list identified in retrospective cohort:

./0_Key_Discoveries/0_Retrospective_Mutations

2. The HBV integration event identified in retrospective cohort:

./0_Key_Discoveries/1_HBV_Event

3. The CNVs identified in retrospective cohort by CONTRA and ichorCNA tools:

./0_Key_Discoveries/2_Segments_from_ichorCNA
./0_Key_Discoveries/3_Gene_annotations_from_CONTRA

Code for the computing of SNVScore, FRAGScore and IFScore

1. SNVScore:

Run ./000_Code_Directory/PREDICT_Code/SNVScore/snv_model.py, using input "SNV_features_retrospective.xlsx"

2. FRAGScore:

Run ./000_Code_Directory/PREDICT_Code/FRAGScore/offtarget_fragscore.py, using bam files as input

3. IFScore:

Run ./000_Code_Directory/PREDICT_Code/IFScore/offtarget_ifs.py, using bam files as input

The computed features for the SNVScore, FRAGScore, IFScore and PREDICT model construction in the retrospective cohort

1. SNVScore:

./1_Model_Features/0_SNVScore_Input_Features

2. FRAGScore:

./1_Model_Features/1_FRAGScore_Input_Features

3. IFScore:

./1_Model_Features/2_IFScore_Input_Features

4. PREDICTScore:

./1_Model_Features/3_PREDICTScore_Input_Features

Code for computing the 12 machine learning performance metrics

Run ./2_ML_Performance_Metrics/0_Code_ML_metrics.r, using input "0_Input"