Proteins are the primary functional units of biology and the direct targets of most drugs, yet there is limited knowledge of the genetic factors determining inter-individual variation in protein levels (protein quantitative trait loci (pQTLs)). Limitations in high-throughput proteomic measurement technology have meant well-powered genome-wide association studies for large number of proteins so far have lagged behind many of the other "omic" studies such as transcriptomics and metabolomics. This is made more challenging by the complexity of human plasma, characterised by high dynamic range spanning several magnitudes of concentrations and a large number of low abundance proteins. By using an expanded high-throughput multiplex aptamer-based proteomic assay with more than twice the proteome coverage of previous studies, I am able to greatly expand on existing knowledge on genetic determinants of human plasma proteins through testing 10.6 million DNA variants against levels of 2,994 proteins in 3,301 individuals. I identify 1,927 genetic associations with 1,478 proteins, replicating many previous associations as well as gaining novel insights into the genetic architecture of the human plasma proteome. I use several approaches to highlight the application of pQTLs to biology and disease. I show several examples linking distant pQTLs to biologically plausible genes and demonstrate the mediation of distant pQTL by local protein levels, highlighting the role of protein-protein interactions. In addition, I find epistatic effects of genetically determined phenotypes (blood group and secretor status) on protein levels. Through linking previous disease associations, I show that disease associated variants are enriched for pQTLs and I provide insights into possible mechanisms underpinning some of the disease loci. Finally, I identify causal roles for protein biomarkers in disease through multivariable Mendelian randomisation (MR) analysis, leveraging on the simultaneous measurement of multiple functionally related proteins in a locus to account for potential pleiotropic effects. Whereas MR studies of plasma proteins have been constrained by availability of few suitable genetic instruments, the data generated here remedy this bottleneck by furnishing an extensive toolkit. Overall, the work within this thesis foreshadows major advances in post-genomic science through increasing application of novel bioassay technologies to major population biobanks.

L’objectif de ce projet était de comprendre comment l’expression du génome influence les caractères de qualité du foie gras, tels que le poids de foie, le taux de fonte et les teneurs, en lipides et protéines, et d’identifier les mécanismes moléculaires sous-jacents. Afin de répondre à ces objectifs, nous avons dans un premier temps étudié l’expression différentielle de protéines selon les phénotypes des foies des canards mulards puis dans un second temps, nous avons entrepris d’identifier des QTL phénotypiques et protéiques (pQTL) à l’aide d’une nouvelle carte génétique composée de marqueurs SNP et microsatellites. Tout d’abord, une optimisation du dispositif expérimental a été entreprise : 3 famille F1, composées de 98 canes backcross et de leurs 294 fils mulards ont été sélectionnés pour leur contribution à des QTL existants liés à la qualité du foie gras. La première approche nous a permis de montrer que les foies ont des profils protéiques et métaboliques différents selon leur phénotype. Ainsi, les foies légers qui fondent peu, avec une faible teneur en lipides mais une forte teneur en protéines présentent un processus anabolique par la surexpression de protéines impliquées dans les métabolismes lipidiques, glucidiques, de synthèse. Au contraire, les foies lourds, fondant beaucoup, avec une forte teneur en lipides mais une faible teneur en protéines présentent des mécanismes de cytoprotection et de réponse au stress. La seconde approche nous a permis de mettre en évidence 30 QTL relatifs à des phénotypes de qualité du foie gras et 50 pQTL relatifs à différentes protéines. Sept chromosomes se démarquent par la ségrégation de plusieurs QTL et pQTL permettant d’émettre des hypothèses quant aux fonctions des gènes sous-jacents à ces QTL. Entre autres, le locus d’APL15 semble lié à la glycolyse et celui d’APL18 à des processus de survie cellulaire. L’ensemble de ces résultats permet ainsi non seulement d’identifier les voies métaboliques impliquées dans la qualité du foie, mais également d’établir un lien entre les caractères, les protéines et les loci des QTL suggérant un déterminisme génétique de ces voies métaboliques impliquées. Ces relations nécessitent d’être approfondies afin de préciser les processus et les gènes impliqués dans la qualité du foie gras. / The aim of this project was to understand how the genome expression influences liver quality traits such as liver weight, melting rate, lipid and protein rates, and to identify the molecular mechanisms underlying it. First, we studied the differential expression of proteins according to liver quality traits of mule ducks. Then we carried out detections of uni-trait and multi-traits phenotypic QTL and protein QTL using a new genetic map containing SNP and microsatellite markers. In preamble to the study, the optimization of the experimental disposal was necessary: 98 backcross dams and their 294 mule sons, composing 3 F1 families were selected because of their contribution to the likelihood of existing QTL related to foie gras quality. The first study showed that livers presented different protein and metabolic profiles according to their phenotypes. Indeed, light livers, with low melting rate, low lipid rate and high protein rate show an over-expression of proteins involved in lipid, glucid or in synthesis metabolism, suggesting an anabolism process. On the contrary, heavy livers, with high melting rate, high lipid rate and low protein rate show cytoprotection and response to stress mechanisms. The second study highlighted 30 QTL related to liver quality traits and 50 pQTL related to different proteins. In particular, 7 chromosomes segregated several QTL and pQTL, permitting to assess hypothesis on the functions of the genes underlying these QTL regions. As an example, the APL15 locus seems linked to glycolysis and the APL18 one seems linked to cell survival ones. All these results helped in identifying metabolic pathways implicated in liver quality as well as establishing a link between traits, proteins and the QTL loci, suggesting a genetic determinism of these pathways. These relationships need to be further studied in order to bring precision to the process and to determine more precisely the genes implicated in the foie gras quality traits.