|
Software for Finding Promoters and First Exons in Human Genome
© 2001 Cold Spring Harbor Laboratory
FirstEF is software developed by Drs. Ramana Davuluri, Ivo Grosse and Michael Zhang at Cold Spring Harbor Laboratory to enable solving the most difficult problem in gene finding – the identification of promoters and first exons in the human genome1. Delineating these elements is essential for establishing gene regulation from functional genomic studies, such as in microarray data analysis. The FirstEF software employs a set of quadratic discriminant functions that recognize such features as CpG islands, promoter regions and first donor sites. To overcome the obstacles of the low signal-to-noise ratio and the heterogeneous nature of the data, the software employs both a decision tree and multiple modeling approaches. Using different models for CpG- and non-CpG related first exons, this software predicts 86% of first exons with 17% false positives. It predicts promoter regions with a sensitivity of 79% and specificity of 54%2. This compares favorably to the respective values, 48% and 43%, for Promoter Inspector3.
Important features of FirstEF include:
(1) the capacity to predict both partially coding and completely non-coding first exons,
(2) the use of distinct models to differentiate between CpG-related and non-CpG-realted first exons and,
(3) an effective model for identifying first donor site.
CSHL is making FirstEF site licenses available to commercial entities for internal use. The license fee for a site-license for FirstEF is $15,000 (executable version).
Davuluri, RV. Grosse, I and Zhang MQ, Computational identification of promoters and first exons in the human genome. Nature Genetics 29:412-417 (2001).
Performance statistics were obtained from the datasets specified in Reference 1, Davuluri et al.. (2001).
Scherf, M., Klingenhoff, A & Werner, T. Highly specific localization of promoter regions in large genomic sequences by PromoterInspector: a novel context.
|
