Gustavo de los Campos

Predictive models of the genetic bases underlying budding yeast fitness in multiple environments

The paper uses machine learning to predict yeast fitness across 35 environments using genetic variants, finding that copy number variants explain most fitness variation and that gene contributions depend on isolate and environment.

Improving Polygenic Score Prediction for Underrepresented Groups Through Transfer Learning

This paper improves polygenic score prediction for underrepresented groups using transfer learning methods implemented in the GPTL R-package, showing better performance than traditional approaches.

Global Genotype by Environment Prediction Competition Reveals That Diverse Modeling Strategies Can Deliver Satisfactory Maize Yield Estimates

The paper reports on a competition to predict maize yield using genetic and environmental data, showing that diverse modeling approaches can achieve good results.

Bayesian Hierarchical Hypothesis Testing in Large-Scale Genome-Wide Association Analysis

This paper introduces a Bayesian Hierarchical Hypothesis Testing method for GWAS that improves power and fine-mapping resolution, leading to more discoveries in serum urate risk variant mapping.

A conserved phenylalanine motif among Teleost fish provides insight for improving electromagnetic perception.

The paper identifies a conserved phenylalanine motif in Teleost fish that may enhance electromagnetic perception, with functional variants linked to freshwater environments.

Local genetic covariance between serum urate and kidney function obtained from local Bayesian regressions.

This paper uses local Bayesian regressions to identify genomic regions with significant genetic covariance between chronic kidney disease and hyperuricemia, finding both novel and previously reported shared loci.

Opportunities and limits of combining microbiome and genome data for complex trait prediction

This paper evaluates the utility of combining microbiome and genome data for complex trait prediction using simulation and variance-component analysis, finding that microbiome data improve prediction accuracy but causal effects are hard to identify.

Regularized selection indices for breeding value prediction using hyper-spectral image data

The paper develops regularized selection indices to improve breeding value prediction using hyper-spectral image data, showing higher accuracy for grain yield than traditional methods.

Deciphering sex-specific genetic architectures using local Bayesian regressions

This paper develops a local Bayesian regression method to detect sex-specific genetic interactions in complex traits, finding many new interactions with small effects.

Analysis of variance when both input and output sets are high-dimensional

The paper introduces two methods for analysis of variance in high-dimensional data and evaluates their performance on genomic datasets.

Transcriptome-based prediction of complex traits in maize

This study predicts complex traits in maize using transcriptome and genetic marker data, finding similar performance between both approaches and highlighting transcriptome's role in capturing trait variation not captured by sequence data.

HaploBlocker: Creation of subgroup specific haplotype blocks and libraries

HaploBlocker identifies subgroup-specific haplotype blocks and constructs libraries to represent genetic variability with fewer parameters than traditional methods.

Imperfect Linkage Disequilibrium Generates Phantom Epistasis (& Perils of Big Data)

The paper shows that imperfect linkage disequilibrium can create false epistasis signals in GWAS, even when the true genetic model is additive, and warns against overinterpreting such results in large datasets.

Quantifying Heterogeneity in the Genetic Architecture of Complex Traits Between Ethnically Diverse Groups using Random Effect Interaction Models

The paper quantifies genetic architecture differences between European-Americans and African-Americans for four traits using random-effect interaction models, finding significant effect heterogeneity.

Accurate Genomic Prediction Of Human Height

The paper constructs genomic predictors for human height, heel bone density, and educational attainment using machine learning, capturing significant variance and resolving part of the missing heritability problem.