Welcome to the hybrid alignment p-value estimation server

Hybrid aligment is a variant of the Smith-Waterman algorithm for similarity detection between biological sequences. The advantage of hybrid alignment over the Smith-Waterman algorithm is that the statistics of hybrid alignments are much better characterized than for the Smith-Waterman algorithm. In order to report a p-value for a Smith-Waterman alignment large simulations are necessary. This server calculates p-values for hybrid alignments with arbitrary scoring matrices and gap costs in realtime. A modified version of the BLAST program that utilizes hybrid alignments instead of Smith-Waterman alignments is available to academic users upon request.

Scoring system

Your email address (mandatory):

Scoring matrix: (unfortunately, you cannot upload your own matrix yet...)

Gap initiation: Gap extension:

Query sequence (you may leave this field empty):

Note, that some combinations of scoring matrices and gap costs do not lie within the local alignment regime and it may thus be impossible to determine p-values !

Parameter setting

You can tune different parameters. Please keep them at their default values if you are not sure about what you are doing.

Estimation of parameter K

Sequence length:
Relative error: %

Estimation of the sequence length correction

Sequence length:
Number of sequences:

If you want to know more about what hybrid alignment is, please read the articles

Yi-Kuo Yu and Terence Hwa, Statistical Significance of Probabilistic Sequence Alignment and Related Local Hidden Markov Models, J. Comp. Biol. 8 (2001), 249-282


Yi-Kuo Yu, Ralf Bundschuh, and Terence Hwa, Hybrid Alignment: High Performance with Universal Statistics, Bioinformatics 18, 864-872 (2002).

For questions and/or comments please send an email to hybparam@bioserv.mps.ohio-state.edu.

9/13/01, Ralf Bundschuh, The Ohio State University