A negative selection heuristic to predict new transcriptional targets

BMC Bioinformatics

Research

Vincenzo Paduano¹  Michele Ceccarelli²  Luigi Cerulo²  Pietro Zoppoli³ 

[1] BioGeM s.c.a r.l., Institute of Genetic Research "Gaetano Salvatore", Ariano Irpino, (AV), Italy;Department of Science, University of Sannio, Benevento, Italy;BioGeM s.c.a r.l., Institute of Genetic Research "Gaetano Salvatore", Ariano Irpino, (AV), Italy;Institute for Cancer Genetics, Columbia University, New York, NY, USA;

关键词: Support Vector Machine;    Negative Selection;    Gene Regulatory Network;    Support Vector Machine Classifier;    Unlabeled Data;   

DOI  :  10.1186/1471-2105-14-S1-S3

来源: Springer

【 摘 要 】

BackgroundSupervised machine learning approaches have been recently adopted in the inference of transcriptional targets from high throughput trascriptomic and proteomic data showing major improvements from with respect to the state of the art of reverse gene regulatory network methods. Beside traditional unsupervised techniques, a supervised classifier learns, from known examples, a function that is able to recognize new relationships for new data. In the context of gene regulatory inference a supervised classifier is coerced to learn from positive and unlabeled examples, as the counter negative examples are unavailable or hard to collect. Such a condition could limit the performance of the classifier especially when the amount of training examples is low.ResultsIn this paper we improve the supervised identification of transcriptional targets by selecting reliable counter negative examples from the unlabeled set. We introduce an heuristic based on the known topology of transcriptional networks that in fact restores the conventional positive/negative training condition and shows a significant improvement of the classification performance. We empirically evaluate the proposed heuristic with the experimental datasets of Escherichia coli and show an example of application in the prediction of BCL6 direct core targets in normal germinal center human B cells obtaining a precision of 60%.ConclusionsThe availability of only positive examples in learning transcriptional relationships negatively affects the performance of supervised classifiers. We show that the selection of reliable negative examples, a practice adopted in text mining approaches, improves the performance of such classifiers opening new perspectives in the identification of new transcriptional targets.

【 授权许可】

CC BY   
© Cerulo et al.; licensee BioMed Central Ltd. 2013