Missing Heritability in GWAS Studies: Enhancing SNP Matrices to Better Track It

Genotype-wide association studies (GWAS) are widely used in our unit, particularly to study plant traits related to adaptation to various stresses. Although plants often yield better results in GWAS than humans, a significant gap remains between the heritability estimated for phenotypes and what classical single-nucleotide polymorphisms (SNPs) can explain: this is referred to as missing heritability.

Genotype matrices are constructed using a single genome as a reference. This framework overlooks presence-absence (P/A) polymorphisms, copy number variations (CNVs), and other structural variants (SVs) that shape the “cloud” and “shell” compartments of pan-genomes. Our results suggest that some of the missing heritability could be revealed by better accounting for insertion/deletion polymorphisms when constructing SNP matrices in order to reduce the number of missing data points. Furthermore, current GWAS rely on biallelic SNP matrices constructed from the most frequent SNP variant—that is, transition variants (A→T or C→G)—even though the degeneracy of the genetic code means that these variations have fewer functional consequences than transversions (A→C or G and T→C or G).

The internship aims to overcome these two limitations by developing highly reusable (containerized) bioinformatics pipelines designed to: 1) for a variable locus—present as an insertion in certain plant lines and as a deletion in others—replace the “NA” coding for a deletion with “missing” coding. 2/ to adapt the MLMM GWAS pipeline—widely used in plants with biallelic SNP matrices—to utilize the currently available but otherwise discarded multi-allelic data. 3/ to test these new pipelines on real-world datasets to evaluate the benefits they provide.

The student will develop these pipelines with support from the GAFL bioinformatics team and the GQMS (Quantitative Genetics and Selection Methodology) team at the GQE-Le Moulon unit. In this context, the student will be based at GAFL in Avignon but will spend one month at the GQE Le Moulon unit (Gif-sur-Yvette). The ideal candidate will have a background in applied statistics or quantitative genetics, be proficient in R, and have a strong interest in bioinformatics.  The student will be co-supervised by:

– Jacques Lagnel, Engineer at GAFL

– Tristan Marie-Huart, Professor at INA-PG

– Nathalie Boissot, Researcher at GAFL

Please send your application or request for additional information to: nathalie.boissot@inrae.fr and jacques.lagnel@inrae.fr