Data sets for variable selection and relation analysis

This data sets are used in the linked publication proposing the two novel approaches mutual forest impact (MFI) and mutual impurity reduction (MIR).

Simulation study 1 was conducted to analyze the bias of importance and relation measures and contains two null scenarios with increasing number of expression possibilities (A) and with increasing minor allele frequencies (B). For each scenario, a classification, regression and survival outcome was simulated. The data contains scripts for simulation and the simulated data.

Simulation study 2 was conducted to analyze the selection of variables in the presence of correlations. The data contains scripts for simulation and the simulated data.

Simulation study 3 was conducted for the comparison of the feature selection approaches under realistic correlation structures. It is based on a realistic covariance matrix (mvn.RData) generated from an RNA-microarray dataset of breast cancer patients with 12,592 genes obtained from The Cancer Genome Atlas. The data contains only scripts for simulation.

The data of the real data application is published in two csv files: "vcf.csv" contains the SNP data of the subset of the plastid genome data set of Solanum Section Petota species (Huang et al., 2019) in a variant calling format file. For this, multiple sequence alignments of 43 genes were conducted with QIAGEN CLC Genomics Workbench 22.0.2 (digitalinsights.qiagen.com) and SNP-sites was subsequently used to generate variant call format (VCF) files. These files were merged into a file of 257 SNPs for further analysis.
"vcf_input_withCountry.csv" contains the same data but with the additional country category. Also the data is in a ready to use format for further analysis.

Further information regarding the dataset can be found in the original publication from Huang et al. (2019) licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/)

Huang, B., Ruess, H., Liang, Q., Colleoni, C., and Spooner, D. M. (2019). Analyses of 202 plastid genomes elucidate the phylogeny of Solanum section Petota. Sci Rep, 9, 4454.