PsRRR - Pathways Sparse Reduced Rank Regression

SGL Power

Pathways sparse reduced rank-regression (PsRRR) is a method for detecting gene pathways associated with a multivariate quantitative trait or phenotype. The software also covers the case of a univariate trait. These methods are based on group lasso penalised regression, and incorporate an adaptive, weight-tuning strategy to reduce bias in pathways selection, and a subsampling procedure to produce a robust measure of pathway importance in finite samples. For a univariate trait, the software also provides the option of simultaneous, pathways-driven SNP selection, for the joint identification of pathways and SNPs associated with the trait of interest. This method is based on the sparse group lasso.

Download:

Python source code

Documentation and licence

Example data set

Sample output

References

Silver M., Janousova E., Hue X., Thompson P. and Montana G. (2012) Identification of gene pathways implicated in Alzheimer's disease using longitudinal imaging phenotypes with sparse regression. Neuroimage, 63(3), Pages 1681-1694

Silver M. and Montana G. (2012) Fast identification of biological pathways associated with a quantitative trait using group lasso with overlaps. Statistical Applications in Genetics and Molecular Biology, vol. 11, issue 1, article 7

 

Back Home