Hi, I am trying to blast against eukaryotic database (nt) but i am confused about what should i consider Bit score or E value? Thanks & Regards, Roy
•
written
by
koustavroy88 •
There is a relatively straightforward answer to this: Both.
In more detail and generally, use E-value for filtering significant hits at a cutoff (e.g. 1E-6 is commonly used). Use bitscore to sort hits, e.g. for extracting the top-ranked hits. While the ordering should be theoretically identical, due to machine number representation, different bitscores can lead to identical E-values, especially for values E-values very close to 0.
A more comprehensive answer would be based on the specific application of the blast results. However, you can take as a rule of thumb.
But note, according to the official definition, the E-value
describes the number of hits one can “expect” to see by chance when searching a database of a particular size
so you need to know the size of the database to use itr properly, or to compare it to other searches.
My suggestion is to never use a bit-score, and don't even know how you came about that idea.
Bit-score is length-dependent, which means that longer sequences have larger bit-scores. You will never get a bit-score of 2000 for a sequence that is 150 residues long. This is to say that a bit-score reflects the relationship, but also the length of a match. We can have a bit score of 250 between two long and unrelated sequences, while the same score between the two shorter sequences will reflect a true relationship.
Hopefully you know what is coming in this section: E-values are length-independent, which is why they reflect the relationships better than bit-scores.
Hi, Actually when i am doing the blast by default it gives the output where i found that its in the order of descending bit score. so i thought is it ok to grab the first hit in the list given in the blast result?
You asked a question and I answered it. Now you are asking something else. It doesn't matter what is used to sort the hits, so if you want to grab a top hit just do it.
Whether that hit is a true homolog or not is a different question. Back to bit-scores and E-values: a top hit that has E=1e-5 or smaller is just about guaranteed to be real. A top hit that has a bit score of 100 may or may not be real.
Log in to answer this question.
