Omics Data Integration Using Kernel Methods (original) (raw)

Kernel-based methods are powerful methods for integrating heterogeneous types of data. mixKernel aims at providing methods to combine kernel for unsupervised exploratory analysis. Different solutions are provided to compute a meta-kernel, in a consensus way or in a way that best preserves the original topology of the data. mixKernel also integrates kernel PCA to visualize similarities between samples in a non linear space and from the multiple source point of view. Functions to select and display important variables are also provided in the package in an unsupervised and kernel association frameworks.

Installation instructions are provided below.

Installation of python dependencies

The following python modules are required for the functions performing feature selection in mixKernel: autograd, scipy, sklearn, numpy

pip3 install autograd
pip3 install scipy
pip3 install sklearn
pip3 install numpy

Installation of Bioconductor dependencies

Two Bioconductor packages are required for mixKernel installation: mixOmics and phyloseq:

install.packages("BiocManager")
BiocManager::install("mixOmics")
BiocManager::install("phyloseq")

mixKernel installation

Finally the installation is completed with:

References

Mariette, J. and Villa-Vialaneix, N. (2018). Unsupervised multiple kernel learning for heterogeneous data integration. Bioinformatics, 34(6), 1009-1015.

Brouard, C., Mariette, J., Flamary, R., & Vialaneix, N. (2022). Feature selection for kernel methods in systems biology. NAR Genomics and Bioinformatics, 4(1), lqac014.