dv-trio

dv-trio performs family-based variant calling by applying DeepVariant to father-mother-child sequencing data to improve detection of single-nucleotide variants (SNVs) and small insertions/deletions (indels) using Mendelian inheritance constraints.

Key Features:

• DeepVariant integration: Utilizes Google's DeepVariant deep neural network to call SNVs and small indels from sequence read pileups.
• Adaptation for trio-based data: Adapts DeepVariant's single-sample approach to accommodate trio (father-mother-child) information for joint analysis.
• Family-based analysis: Incorporates genetic data from trios to enhance detection of inherited and de novo variants relevant to disease mutation discovery.
• Image-based deep learning: Converts sequencing read pileups into image representations to enable application of convolutional deep learning models.
• Mendelian genetic model integration: Applies Mendelian inheritance constraints to refine and validate variant calls within the trio context.

Scientific Applications:

• Disease Mutation Discovery: Analyzes trio sequencing data to improve identification of disease-associated inherited and de novo mutations for genetic research and diagnostics.
• Enhanced Variant Calling Accuracy: Uses family-based and Mendelian information to achieve more precise variant calls compared with single-sample approaches.

Methodology:

Converts sequence read pileups into images, applies DeepVariant's deep neural network to detect SNVs and small indels, and integrates Mendelian inheritance constraints from father-mother-child trios to refine and validate variant calls.