Science Cover Art

The following are journal covers I designed to promote two genetics papers as a research assistant with Dr. Quan Long at the University of Calgary.

Cover submission for SplitTWAS

Published as cover of Genetics: Volume 220, Issue 2, February 2022.

A statue of a human figure and a geometric statue stand in front of an arched bridge. The bridge is reflected in water to form a double helix shape.
This artwork represents the central idea of the paper “Disentangling genetic feature selection and aggregation in transcriptome-wide association studies”, which splits the central model in transcriptome-wide association studies (TWAS) into two steps: 1) using gene expressions to select genotype features, and 2) using phenotypes to score aggregated genotypes. Although there is a causal relation between genotype and expression, this is not equivalent to the genetically regulated gene expression (GReX) calculated by traditional TWAS. GReX is a statistical association, and can either indicate that genotype predicts expression (the traditional TWAS view), or that expression selects relevant genotypes. The latter perspective allows GReX to be disentangled into two separate models for selecting and aggregating features (genotypes). In this view, genotypes switch place with expressions to become the bridge (center) that links gene expression and phenotype (statues) via statistical association.

Cover submission for PoolHapX

The output of a paper shredder is pieced back together into sheets of paper by a fantastical contraption labelled PoolHapX. The machine has four parts labelled graph color, AEM, BFS, and regression. The shredded paper creates a DNA shape.
This artwork illustrates the software described in the paper “Reconstruction of microbial haplotypes by integration of statistical and physical linkage in scaffolding”, which recovers the genetic sequences and prevalence of multiple subpopulations of organisms (haplotypes) from pooled DNA sequencing data. The software has four steps: 1) graph coloring to reduce the set of possible sequences, 2) hierarchical Approximate Expectation Maximization to identify likely subsequences, 3) breadth-first tree search to enumerate potential haplotypes, and 4) regularized regression to estimate haplotype frequencies from the observed frequencies of variants and variant pairs.

2021 November 28