A Review of Matched-pairs Feature Selection Methods for Gene Expression Data Analysis

This code is for this paper:

• Sen Liang, Anjun Ma, Sen Yang, Yan Wang, Qin Ma, "A Review of Matched-pairs Feature Selection Methods for Gene Expression Data Analysis”, Computational and Structural Biotechnology Journal, 2018, 16:88-97. https://doi.org/10.1016/j.csbj.2018.02.005.

In this paper, we compare the performance of 10 feature-selection methods (eight MPFS methods and two traditional unpaired methods) on two real datasets by applied three classification methods, and analyze the algorithm complexity of these methods through the running of their programs.

Try use this code

S1-ttest.R

• Paired t-test
• Modified paired t-test
• Fold-change paired t-test

S22-boosting.R (copy from orignial paper code)

• Boosting Weighted L2 Loss (WL2Boost).
• 1-Step Penalized Quasi-Likelihood (1-Step PQLBoost)

S3-CLR.R

• Random penalized conditional logistic regression (RPCLR)
• Penalized Conditional and Unconditional Logistic Regression (PCU-CLR)

3rdpart / BVS-CLR folder

• Bayesian Variable Selection Conditional Logistic Regression (BVS- CLR)

3rdpart / MRMD-master.zip

• MRMD algorithm

MRMR algorithm

• refer to orignal paper

Compare experiments

try use run.R script