Bayesian Semi-Nonnegative Matrix Tri-Factorization
BSNMTF is a semi-nonnegative matrix tri-factorization method in the framework of Bayesian learning to identify pathways associated with cancer phenotypes (e.g., cancer sub-types or treatment outcome) from a real-valued input matrix (e.g., copy number alterations or normalized/log transformed gene expression data). In contrast to non-negative factorization methods, which are limited to non-negative input only, our method takes a real-valued matrix as input and is able to find the up/down regulated patterns associated with the pathways from the data. We also place structured spike-and-slab priors (which are encoded with the pathways and a gene-gene interaction (GGI) network) on the centroid matrix so that even a set of genes that is not initially contained in the pathways (due to the incompleteness of the current pathway database) can be involved in the factorization in a stochastic way specifically, if those genes are connected to the member genes of the pathways on the GGI network.
-
BSNMTF_supplementary_PSB2020.pdf (supplementary material): contains detailed derivations of the variational update rules and additional information
-
BSNMTF_top3_allcombined_geneset_enrichment.xlsx: cotains the pathway analysis in the main paper
-
MATLAB codes (./matlab_codes)
3-1 bsnmtf.m (Bayesian Semi-Nonnegative Matrix Tri-Factorization): the main Matlab function
3-2 script_BSNMTF_simulation_Z0_incomplete.m: the experiment on the simulation data in the main paper
3-3 script_BSNTMF_TCGA_STAD_Subtypes.m: the experiment on the TCGA gastric cancer dataset in the main paper
3-4 script_BSNTMF_ImmunotherapyResponse.m: the experiment on the metastatic gastric cancer immunotherapy clinical-trial dataset in the main paper
3-5 script_BSNMTF_supp_experiments.m: the experiments on the simulation dataset in the supplementary material
To test BSNMTF (on the simulation data in the main paper), run script_simulation_Z0_incomplete.m (Matlab or Octave, but we have tested our code only on Matlab)
call bsnmtf.m (input arguments: X, U, Z0, A, options)
To rune the codes, you need the following optimziation toolbox: Limited memory BFGS: https://github.com/stephenbeckr/L-BFGS-B-C
The data files in the script files, script_BSNTMF_TCGA_STAD_Subtypes.m and script_BSNTMF_ImmunotherapyResponse.m, are not included in this repository. Please contact the authors if you want to run both scripts.
This work is included in the Proceedings of Pacific Symposium on Biocomputing (PSB 2020), https://psb.stanford.edu/psb-online/proceedings/psb20/.
Contact information: Sunho Park (parks[nospam]@ccf.org)