BioSysBio:abstracts/2007/Martino Barenco

=rHVDM – A fast and user-friendly R package to predict transcription factor targets from microarray time series data.= Author(s): Martino Barenco (1,5), Crispin Miller(2), Sonia Shah (3), Daniel Brewer (4,5), Robin Callard(1,5), Jaroslav Stark(6,5), Michael Hubank(1,5). Affiliations:(1) Institute of Child Health, UCL (2) Paterson Institute of Cancer Research, Manchester (3) Bloomsbury Centre for Bioinformatics, UCL (4) Institute of Cancer Research, Sutton (5) CoMPLEX, UCL (6) Department of Mathematics, Imperial College Contact:email: m dot barenco at ucl dot ac dot uk Keywords: 'time course' 'gene expression' 'Dynamic model' 'gene regulation'

Abstract
Researchers dealing with gene microarray data are faced with daunting quantities of data in which lie hidden important information including transcription factor activity profiles. We developed a technique, HVDM (Hidden Variable Dynamic Modelling)(1), which uses data from a small training set of known transcription factor (TF) targets to deduce the activity profile of the transcription factor - the hidden variable in the system. Other targets of the same TF can then be identified by running the model on microarray time series data plus a single anchoring degradation measurement. The sampling rate can be irregular, replicates are not required and measurement errors are explicitly taken into account so that results are ranked according to confidence. We have now generated a Bioconductor version of HVDM. rHVDM is an R package which uses fast algorithms, including gradient-based optimisation and a linear simplification of the underlying ODE model. As a result, a thousand genes can be screened in about the time it takes to make a cup of tea.