Branch and bound computation of exact p-values
Open Access
- 7 August 2006
- journal article
- research article
- Published by Oxford University Press (OUP) in Bioinformatics
- Vol. 22 (17) , 2158-2159
- https://doi.org/10.1093/bioinformatics/btl357
Abstract
Summary: P-value computation is often used in bioinformatics to quantify the surprise, or significance, associated with a given observation. An implementation is provided that computes the exact p-value associated with any observed sample, against a null multinomial distribution, using the likelihood-ratio statistic. The efficient branch and bound code, far exceeding the full enumeration implemented by commercial packages, is especially useful with small sample, sparse data and rare events, common scenarios in bioinformatics, where approximations are often inaccurate and inappropriate. This code base can also be adapted to compute exact p-values of other statistics in diverse sampling scenarios. Availability: Freely available at Contact:jill@soe.ucsc.edujillKeywords
This publication has 4 references indexed in Scilit:
- Efficient Exact p-Value Computation for Small Sample, Sparse, and Surprising Categorical DataJournal of Computational Biology, 2004
- Dynamic Programming Algorithms for Two Statistical Problems in Computational BiologyPublished by Springer Nature ,2003
- Exact inference for categorical data: recent advances and continuing controversiesStatistics in Medicine, 2001
- Goodness-of-Fit Statistics for Discrete Multivariate DataPublished by Springer Nature ,1988