Sequence vs. saHMM saHMM vs. Sequence DB About Help Contact

FAQ - Frequently asked questions


What does "FISH" stand for?

"FISH" - Family Identification with Structure Anchored Hidden Markov Models


How do I cite the FISH server?

"FISH - Family Identification of Remote Sequence Homologues using Structure Anchored Hidden Markov Models" (Nucleic Acids Res. 2006 Jul 1;34(Web Server issue):W10-4.)


What is a good hit?

A "good hit" is a hit with an E-value as close to 0 as possible. In our tests, an E-value cut-off of 0.1 gives a coverage of about 83%, i.e. about 83% of the family members are found by the saHMMs in our tests. The accuracy of the hits is 98.6%, meaning that only 1.4% of the hits obtained are to sequences from the incorrect family. If we consider the superfamily level, 99.9% of the hits are correct.

Considering only the top hit and an E-value below or equal to 0.1 in a search with sequence vs. the saHMM-db, then as many as 99.3% of the hits are to correct saHMMs. With this in mind we suggest the user to trust hits with E-values below 0.1, especially the top hits. However, also hits with higher E-values might be correct but should be considered with caution.


How do I interpret the ´+´s in the alignment to the saHMM consensus?

The pairwise alignments consist of three lines. First is the query sequence or, in the case of search with saHMM to sequence database, the sequence hit. The middle line shows letters for matches to the highest probability amino acids in the saHMM or a ´+´ when the match in this position has a positive score. The third line is the HMM consensus.


What is the saHMM consensus?

The saHMM consensus is the most likely sequence according to the saHMM. The amino acid displayed in each consensus position is the one with the highest probability at that position according to the saHMM. Capital letters are residues with a probability > 0.5.


How are the pairwise sequence identities calculated?

The identity displayed with a pairwise alignment is the identity of the two sequences aligned, according to the alignment. All pairwise alignments are extracted from a multiple sequence alignment derived using a saHMM.


How are the pairwise alignments constructed?

The pairwise alignments are extracted from the multiple alignment of the query sequence/-s and the structure anchored alignment used as the basis for the saHMM, keeping the saHMM anchor.

Site Map
Site Map
Copyright © 2006 UCMP. All rights reserved.