Transcriptional network inference from functional similarity and expression data: a global supervised approach

Area of Science:

• Systems Biology
• Bioinformatics
• Computational Biology

Background:

• Inferring biological networks, particularly gene transcriptional regulatory networks, from postgenomic data is a significant challenge in systems biology.
• Existing reverse engineering algorithms for transcriptional networks often exhibit limited predictive performance.
• Understanding transcription factor (TF)-target gene interactions is crucial for deciphering gene regulation.

Purpose of the Study:

• To introduce a novel method, TNIFSED (Transcriptional Network Inference from Functional Similarity and Expression Data), for inferring gene transcriptional regulatory networks.
• To evaluate the performance of TNIFSED by comparing it with existing state-of-the-art methods.
• To assess the suitability of TNIFSED for identifying target genes of transcription factors, including those with few known targets.

Main Methods:

• TNIFSED integrates gene expression profile correlations, partial correlations, and gene functional similarities.
• A supervised classifier is employed within the TNIFSED framework.
• The method was applied to predict transcriptional networks in Escherichia coli and Saccharomyces cerevisiae using large-scale Affymetrix array datasets.

Main Results:

• TNIFSED demonstrated superior predictive performance compared to unsupervised state-of-the-art methods, as indicated by receiver operating characteristics (ROC) and F-measure.
• The method showed comparable performance to the supervised SIRENE algorithm.
• TNIFSED excelled in identifying target genes for transcription factors with a limited number of known targets, often referred to as "orphan" TFs.

Conclusions:

• TNIFSED offers an effective approach for transcriptional network inference by integrating diverse data types.
• The method is particularly valuable for discovering target genes of "orphan" transcription factors.
• TNIFSED complements existing algorithms like SIRENE, providing a valuable tool for systems biology research.