Protocol Online logo
Top : Forum Archives: : Bioinformatics and Biostatistics

What is e value? - (Mar/03/2005 )

How do you interepret the e value for a sequence alignment? Are there any other parameters that can be employed to asses the significance level of the alignment?

Thank you in advacne for you kind reply

-sharath-

the e value give a measure of the similarity of sequences, the lower the e value, the higher the congruity of your query sequence and the retrieved sequence. e values of 0 mean that there's an exact match for you sequence here...

mike

-jadefalcon-

e-values in blast results represent the probability of the alignment occurring by chance. It is a statistical calculation based on the quality of alignment (the score) and the size of the database. For example if an alignment obtained from one database has an evalue of x, the exact same alignment obtained from a database of different size will have an evalue of y.

An evalue of 1e-3 is saying that there is a 0.001 chance that that alignment would exist in the database by chance, that is, if the database contains 10000 sequences, then you might expect that alignment to occur maybe 10 times.

An evalue of 0 is actually a rounded down probability (maybe 1e-250 or something), and is simply saying that there is (almost) no chance that alignment can occur by chance.

The score is the measure of similarity between two sequences, and is calculated from the alignment matrix.

D.

-dlaw-

yes dlaw's correct here, but then I'm just a biologist, not a computer person biggrin.gif !
I just wanted to give a very simplified explanation, so maybe this one was simplified to the point where is was, strictly speaking, wrong...

mike

-jadefalcon-