Output details

Effect of using varying negative examples in transcription factor binding site predictions

<24> This paper was selected as one of the two best papers at the conference EvoBio, Toulouse 2011. It represents a novel use of negative examples. Along with the novel cross-validation method, our new procedure is a large improvement when dealing with this type of biological data. We are able to predict, with great accuracy, the location of important regulatory regions of the genome. This performance is much better than anything else that has been reported in the literature. The work led to a series of follow-up studies in collaboration with the Sanger Institute (Hinxton, Cambridge).