A test for the statistical significance of DNA sequence similarities for application in databank searches |
| |
Authors: | Mott, R. F. Kirkwood, T. B. L. Curnow, R. N. |
| |
Affiliation: | Laboratory of Mathematical Biology, National Institute for Medical Research Mill Hill, London NW7 IAA 1Department of Applied Statistics, University of Reading Reading RG6 2AN, UK |
| |
Abstract: | A method is developed, based on word-searching, which providesa rapid test for the statistical significance of DNA sequencesimilarities for use in databank searching. The method makesallowance for the lengths and dinucleotide compositions of thesequences being compared. A way is also described to calculatethe power of the test, i.e. the probability of detecting a givensimilarity as being statistically significant. The effects onthe power of the test of the scoring method, word length, sequencelength, and sequence composition are examined. A novel scoringmethod is shown to be superior to the method currently usedin most word-searching algorithms. Received on August 3, 1988; accepted on December 12, 1988 |
| |
Keywords: | |
本文献已被 Oxford 等数据库收录! |
|