Challenges in proteogenomics: a comparison of analysis methods with the case study of the DREAM proteogenomics sub-challenge

Table 1 Contrasting Features of the Methods

Model	Nonlinear interactions	Computationally efficient	Probabilistic model
LASSO		x
Random Forests	x	x
Fuzzy Logic	x		x
Bayesian Networks	x		x

Contrasting features of the different algorithms used to predict protein levels from transcripts. “Nonlinear interactions” indicates that the algorithm does not assume that protein levels are a linear function of transcript levels. “Computationally efficient” means that predictions were able to be made in less than 12 h on the Ohio Supercomputer Center (OSC) cluster using all transcripts in the data as input. “Probabilistic model” means that the predictions are given as probabilities, representing uncertainty in the data

ISSN: 1471-2105