Deep learning models that classify overlaps produced by assembly algorithms like Raven.
Data directory contains examples of training data.
usage: main.py [-h] [--dataset DATASET] [--learning_rate] [--cuda] [--type TYPE] [--extended_cigar] [--fn]
optional arguments:
-h, --help show this help message and exit
--dataset DATASET train dataset file
--learning_rate learning rate for Adam (default: 0.001)
--cuda use CUDA for model training
--type TYPE Model type: A, B or C
--extended_cigar Use extended cigar strings with X and = instead of M
--fn Classify false negatives
When using extended_cigar option, dataset must have = and X for matches, for example:
python3 main.py --dataset data/example.txt --fn --extended_cigar
whereas if extended_cigar flag is not passed, dataset must contain only M for matches:
python3 main.py --dataset data/exampleM.txt --fn
fn option classifies true positives, false positives and false negatives, while without the option model classifies only in true and false positive class.