R package ipflasso: Integrative Lasso with Penalty Factors. The core of the package is cvr2.ipflasso(), an extension of glmnet to be used when the (large) set of available predictors is partitioned into several modalities which potentially differ with respect to their information content in terms of prediction. For example, in biomedical applications patient outcome such as survival time or response to therapy may have to be predicted based on, say, mRNA data, miRNA data, methylation data, CNV data, clinical data, etc. The clinical predictors are on average often much more important for outcome prediction than the mRNA data. The ipflasso method takes this problem into account by using different penalty parameters for predictors from different modalities. The ratio between the different penalty parameters can be chosen from a set of optional candidates by cross-validation or alternatively generated from the input data.
Keywords for this software
References in zbMATH (referenced in 3 articles , 1 standard article )
Showing results 1 to 3 of 3.
- Rauschenberger, Armin; Ciocănea-Teodorescu, Iuliana; Jonker, Marianne A.; Menezes, Renée X.; van de Wiel, Mark A.: Sparse classification with paired covariates (2020)
- De Wiel, Mark A. van; Te Beest, Dennis E.; Münch, Magnus M.: Learning from a lot: empirical Bayes for high-dimensional model-based prediction (2019)
- Boulesteix, Anne-Laure; De Bin, Riccardo; Jiang, Xiaoyu; Fuchs, Mathias: IPF-LASSO: integrative (L_1)-penalized regression with penalty factors for prediction based on multi-omics data (2017)