Code of Canid Diversity: Identifying Mutations Driving Sheepdog Behavior Using Deep Learning

Overview

The vast number of mutations within a genome makes it a central challenge in genomics to determine the process behind how mutations result in phenotypic diversity. Humans have bred dogs over thousands of years,1 making them the most phenotypically diverse mammalian species and an unparalleled model to study.2 Dutrow et al. researchers have used genome-wide association studies (GWAS) that show the genetic basis of dog lineages to associate genetic variants with lineage diversification. 1,3 Since variants occur in ancestrally correlated variant blocks, it can be difficult to isolate functional variants. To overcome these challenges, I trained a deep learning model that predicts variant function across 29 canine brain cell types. I hypothesized that few non-coding variants are functional in causing sheepdog lineage-specific phenotypes and can be detected by a trained convolutional neural network model. This model identified 2259 functional variants out of the 38,107 sheepdog lineage correlated variants from the Dutrow et al. GWAS. I found that most lineage-associated variants did not cause a functionally significant change. Significant sheepdog variants were found to be closer to gene transcription start sites. Also, Axon gene-associated variants caused increased changes in neuronal cell types. The fine-map study results show that only 5.928% of the over 38 thousand sheepdog variants are functionally causal, meaning they do not cause gene regulatory changes by measuring chromatin accessibility, opening or closing chromatin.4,5 Ultimately, the results support the hypothesis that functional phenotype-driving variants are detectable. The fine-mapped variants that are isolated help solidify the relationship between genotype and sheepdog phenotype, providing the foundation for future developments, such as drugs that can cure diseases by targeting specific causal variants.

Competition history

• CSEF 2026 Biochemistry/ Molecular Biology (Senior Division) · Entry S-04-33

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

• All Cellular and Molecular Biology projects
• CSEF Cellular and Molecular Biology projects
• Every CSEF 2026 project