SUPR
AI methods for identifying genetic associations and for prediction modelling
Dnr:

sens2021565

Type:

SNIC SENS

Principal Investigator:

├ůsa Johansson

Affiliation:

Uppsala universitet

Start Date:

2021-07-01

End Date:

2023-07-01

Primary Classification:

30107: Medical Genetics

Allocation

  • Castor /proj at UPPMAX: 100 GiB
  • Castor /proj/nobackup at UPPMAX: 100 GiB
  • Cygnus /proj at UPPMAX: 100 GiB
  • Cygnus /proj/nobackup at UPPMAX: 100 GiB
  • Bianca at UPPMAX: 2 x 1000 core-h/month

Abstract

In this project we will use Artificial Intelligence (AI) tools with whole-genome sequencing (WGS) data to: 1) Identify novel gene-phenotype associations 2) Use genetic data to predict phenotypes In genetic studies, traditionally a GWAS approach has been used, where each genetic variant is moulded separately in relation to its effect on a phenotype. With AI methods, many genetic variants can be investigated jointly, which allows for detection of non-linear and non-additive effects. In this project we will test the performance of different traditional ML and AI methods in relation to a large set of quantitative phenotypes but also in relation to simulated phenotypes