DeepConflation based CNN model is used to generate alignment of fasta file . Model try to match a query string in a referance genome to provide location of match.
Model uses CNN to balance noise in query data and match with appropiate referance genome string. To train model a simulated training data is generated by applying mutation and indel to referance gemone and then data is converted to images with refDataSetCreator.py file. The DataGenerator File create training data by reading BAM file and convert it to hot vector encoding of fix length and corresponding fix reference string with querry position and matched referance postion. Model is trained using genome_cnn_train.py For more details please refer to paper DAVI https://iopscience.iop.org/article/10.1088/2632-2153/ab7e19/pdf