DeepSom Neural Network Classifier for calling somatic SNP and INDEL variants without a matched normal.
The Neural Network Classifier (NNC) can be used in three modes:
-
Train mode. During training, the NNC learns relationships between variants and variant classes.
-
Evaluation (test) mode. The performance of the trained model is evaluated using an independent test set. Such metrics as ROC AUC or test loss can be calculated.
-
Inference mode. The probability for a given variant to be somatic is computed based on the NNC output.
See dataprep on how to prepare data for the NNC.
See NNC on how to run the NNC.