Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: Master's Award: Canada Graduate Scholarships

Université de Sherbrooke

$17,500.00

Deep learning based annotation of functional pseudogenes

EF 2022-23

Résumé
Thousands of functional coding sequences (CDS) are missing from human reference genome annotations. These CDS, often referred to as alternative ORF, have eluded annotations because of their small size or unexpected loci. Despite recent efforts from the scientific community to integrate these robustly detected CDS to annotations, a category remains systematically discarded from analyzes: CDS located within pseudogenes. Pseudogenes are identified as any genomic sequence similar to a coding gene (parental gene) but that is defective and often without introns. However, these criteria are ill-advised to predict the functionality of a gene or absence thereof. In human, 9% of functional genes in human lack introns, and the presence of mutations doesn’t always amount to loss of function (ex. UBBP4, NOTCH2NL). This dubious definition as non-functional elements results in their exclusion from omics and fundamental research analyzes. Thus, there is a real need to evaluate the functional character of pseudogenes in a systematic and unbiased manner to elucidate their role in biological processes. My project aims to build and train a deep learning algorithm allowing the prediction of CDS within pseudogenes. I will use the OpenProt proteogenomics resource, co-developed by the laboratory of Pre Brunet, to establish a set of confident CDS. The neural network architecture draws inspiration from the U-Net convolutional network (CNN), an algorithm optimized for image segmentation. I will adapt this architecture to take advantage of its optimized performance in recognizing features far apart in space. This project will allow the annotations of hundreds of CDS to reference annotations and will contribute to recognizing pseudogenization as an active mechanism of genome evolution.
Programme
Master's Award: Canada Graduate Scholarships
Emplacement
Quebec, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
CIHR Funding Decisions Database

Tel que publié

Id
486302
Name
Lapointe Xavier
Region
Québec
Country
Canada
Orgtype
Institution Paid; Research Institution
Prcname2
Special Cases - Awards Programs
Projecttitle
Deep learning based annotation of functional pseudogenes
Programtype2
Studentship Programs
Orgnameinp2
Université de Sherbrooke
Cihramount2
$17,500
Cihrequipment2
$0
Programname2
Master's Award: Canada Graduate Scholarships
Keyworddelim
Bio-Informatique; Génome; Génétique; Ia; Omiques; Pseudogènes
Abstract
Thousands of functional coding sequences (CDS) are missing from human reference genome annotations. These CDS, often referred to as alternative ORF, have eluded annotations because of their small size or unexpected loci. Despite recent efforts from the scientific community to integrate these robustly detected CDS to annotations, a category remains systematically discarded from analyzes: CDS located within pseudogenes. Pseudogenes are identified as any genomic sequence similar to a coding gene (parental gene) but that is defective and often without introns. However, these criteria are ill-advised to predict the functionality of a gene or absence thereof. In human, 9% of functional genes in human lack introns, and the presence of mutations doesn’t always amount to loss of function (ex. UBBP4, NOTCH2NL). This dubious definition as non-functional elements results in their exclusion from omics and fundamental research analyzes. Thus, there is a real need to evaluate the functional character of pseudogenes in a systematic and unbiased manner to elucidate their role in biological processes. My project aims to build and train a deep learning algorithm allowing the prediction of CDS within pseudogenes. I will use the OpenProt proteogenomics resource, co-developed by the laboratory of Pre Brunet, to establish a set of confident CDS. The neural network architecture draws inspiration from the U-Net convolutional network (CNN), an algorithm optimized for image segmentation. I will adapt this architecture to take advantage of its optimized performance in recognizing features far apart in space. This project will allow the annotations of hundreds of CDS to reference annotations and will contribute to recognizing pseudogenization as an active mechanism of genome evolution.
Competitiondate
202212
Approvedterm2
1 yr 0 mth
Pinamesdelim
Lapointe, Xavier
Approvedterm
1 an 0 mois
Cihrcontribution2
~ $17,500~ $0~