google/deepvariant
DeepVariant uses a convolutional neural network trained on TensorFlow to call genetic variants from DNA sequencing data.

DeepVariant is an analysis pipeline that takes aligned sequencing reads (BAM/CRAM), converts them into pileup image tensors, and classifies each tensor using a deep convolutional neural network to identify genetic variants. The pipeline outputs results in standard VCF or gVCF format. It supports multiple sequencing technologies including Illumina, PacBio HiFi, Oxford Nanopore, Complete Genomics, and Roche SBX, and can perform hybrid calling combining multiple data types.