The need for developing epistasis detection methods has become evident and more urgent over the past years. While GWAS have been able to explain part of the heritability present in various phenotypes, for complex diseases such as ALS, cancer and Alzheimer’s a major part of heritability remains unexplained, e.g. only 8.5% of heritability in ALS is explained by GWAS. This is likely (partially) due to non-linear interactions that GWAS cannot catch.
Over the past decade researchers from several fields including bioinformatics, computer science, statistics and mathematics have turned to developing data analysis methods to detect epistatic disease-causing interactions from large genome data. Various techniques are currently investigated, such as Bayesian approaches, combinatorial methods, exhaustive search, machine learning, deep learning and more.
Researchers are working on this topic coming from different backgrounds and taking various approaches. This workshop aims to break the silos and facilitate discussion and collaboration between specialists who are approaching epistasis with different methodologies. We hypothesise that each approach is ideally fitted to specific circumstances and that an ideal approach would be a hybrid of different methods. We think that currently researchers in this field are not aware of the specific circumstances ideally suited to methods that they do not already use regularly; even if they are aware of the strengths and weaknesses of their favoured approach. In this workshop we aim to identify the benefits and limitations of the various approaches both theoretically and experimentally.
17 – 21 July 2023 at Lorentz Center@Oort