PrediXcan and TWAS methods in general correlate genetically predicted levels of gene expression traits with complex traits to understand the mechanism behind GWAS loci. A key component is the training of gene expression traits. A tutorial on how to generate elastic net models can be found in this link

Elastic net is a good all purpose prediction approach for complex traits and has been shown to perform well for gene expression traits. Depending on your goals, you may want to use a different approach. For example, if the goal is to maximize the reliability (low false positive) of putatively causal genes, then we showed that a method that uses genetic variants more likely to be causal may work better. Explained in this paper. In the GTEx GWAS subgroup we chose the models that are based on fine-mapping, called mashr-based.


Text and figures are licensed under Creative Commons Attribution CC BY 4.0. The source code is licensed under MIT.

Suggest changes

If you find any mistakes (including typos) or want to suggest changes, please feel free to edit the source file of this page on Github and create a pull request.


For attribution, please cite this work as

Haky Im (2021). Training Gene Expression Prediction Models. ImLab Notes. /post/2021/07/09/training-gene-expression-prediction-models/

BibTeX citation

  title = "Training Gene Expression Prediction Models",
  author = "Haky Im",
  year = "2021",
  journal = "ImLab Notes",
  note = "/post/2021/07/09/training-gene-expression-prediction-models/"