BLAST E-values

Hi everyone,

I have a BLAST related question:

When you get a really good match the E value is reported as 0.0, presumably because it is actually so low it is below a threshold level that will be reported.  It seems a bit confusing to report E values as 0.0 though so I was wondering if anyone knows what the cut-off is?

Cheers,

Amy

3 responses to “BLAST E-values

  1. Deborah Alongi Johnson

    I think the cutoff is E-200. An E value of 0.0 usually represents an exact or near exact match, where the probability of a random match is essentially 0. I’ve reported E values of 0.0 in a publication before.

    Deborah

  2. Hi Amy (et al),

    A description of E values can be found here: http://www.ncbi.nlm.nih.gov/BLAST/blastcgihelp.shtml#expect

    The key point is that the lower the E value, the more likely it is that you have a match. People often take a threshold of e-10 (i.e 0.0000000001) or lower as evidence of a good match. If the match is really, really good then a value of 0.0 will be returned. I think that E values which are lower than about e-150 or e-200 tend to get rounded down to zero. In practical terms though, it doesn’t matter much whether your hit is e-100, e-150 or 0.0. They are all very very good matches that could not have arisen by chance.

    The actual E value is a function of the number of matches and mismatches, the length of the alignment and (I think) the size of the database you are searching against.

  3. Thanks Jon. I emailed the BLAST help people and they told me that any Expect value less than e-179 is rounded down to zero. So I guess it is better to report these E values as <e-179?

Leave a Reply