Inferring transcription factors protein activities by combining binding information via Gaussian process regression

Abstract EN

The most basic step in understanding gene regulated is performed by identifying the target genes regulated by transcription factors (TFs) Proteins.

Protein is produced by Transcription factors Proteins that promote or repress transcription of other genes; they play a very important role in gene networking and affecting for occurring the disease.

The analysis of gene expression of time series underpins various biological studies.

This work has focused on the difference in transcriptional regulation between two strains of mice.

The mice were considered in two forms Wild type SOD1-G93A and Ntg mutations (SOD1 is a transcription factor Protein that induces ALS).

The data interest because the phenotype of the two mutant strains differs.

One of the strains succumbs to ALS far quicker than the other; we suggested a model to infer Transcription Factor Proteins Activities and correlated with genes targeted.

We build Gaussian process with particular covariance function for reconstructing transcription factor activities given gene expression profiles and a connectivity matrix, and we introduce a computational trick, based on Singular Value Decomposition (SVD) to enable us to efficiently fit the Gaussian process in a reduce ’TF activity ’ space.

Performing the basic step in understanding regulated genes is identifying these genes by transcription factors.

Gaussian processes offer an attractive trade-off between usability and efficiency for the analysis of microarray time series.

The Gaussian process framework with Coregionalization model offer a natural way of handling biological replicates and correlated output and inferred the activity of Transcription factors Proteins for four cases the genes alter its behavior, we proved the significates TF using DAVID to analysis pathway.