DeepVariant

DeepVariant is an analysis pipeline that uses a deep neural network to call genetic variants from next-generation DNA sequencing data.

Key Capabilities

• High Accuracy Variant Calling: Significantly outperforms traditional statistical tools (like GATK) for identifying single nucleotide polymorphisms (SNPs) and small insertions and deletions (indels), especially in difficult-to-sequence genomic regions.
• Image-Based Approach: Innovatively treats variant calling as an image classification problem, visualizing aligned reads as “pileup images” and running them through a convolutional neural network (Inception v3).
• Broad Platform Support: Models are trained and highly optimized for various sequencing technologies, including Illumina (short-read), PacBio (HiFi long-read), and Oxford Nanopore.

Usage in Healthcare

Accurate variant calling is the foundational step for all clinical genomics, directly impacting the diagnosis of rare mendelian diseases and targeted cancer therapies. DeepVariant provides clinical labs and bioinformaticians with state-of-the-art accuracy that can be deployed on-premises to maintain strict data privacy compliance for patient genomes.