Semi-supervised mixture models with the lcmix R package for integrative genomics analysis
Semi-supervised and unsupervised mixture models on either sequence-derived features only (1A) or all features (1B). Train models on pre-2002 essential genes (+ training examples only, n = 64) and test on post-2002 essential and non-essential genes (+/- examples). Each script plots an ROC curve comparing the semi-supervised mixture model to the unsupervised mixture model.
Semi-supervised and unsupervised mixture models on either sequence-derived features only (2A) or all features (2B). Train models on essential genes (+ training examples only, varying number of training examples) and test on remaining essential and non-essential genes (+/- examples). Each script plots an ROC curve comparing semi-supervised mixture models and unsupervised mixture models at varying numbers of training examples.
Semi-supervised and unsupervised mixture models on either sequence-derived features only (3A) or all features (3B). Train models on essential and non-essential genes (+/- training examples, varying number) and test on remaining essential and non-essential genes (+/- examples). Each script plots an ROC curve comparing semi-supervised mixture models and unsupervised mixture models at varying numbers of training examples.
Loads required packages and some lcmix helper functions.
Includes wrapper functions for the modeling and plotting used in all the scripts above.