BLAST with 2 different databases

[tickteng]

Hi, I have some sequence results that I blast to 2 different databases (one is NCBI's nt and one other local databases), the result comes out a bit differently. The hit were similar, but the number of read that hit the sequence were somehow different.

Is there any way I can do some statistical analysis to decide which database is better? Or is there any other way to decide.

[r.rosati]

I confess I'm not sure I fully understood your question. Do you mean that there are certain hits that are retrieved from one database, and don't appear when searching on the other? If so, then it might be related to E (the Expect value). Since E is the chance of retrieving a similar hit by random, databases of different size (total number of bases) will return hits with different E values: the smaller the database, the smaller the E. So, you might not retrieve hits from the big database if the E is just big enough to be above your cut-off value, while you might still retrieve that hit from a smaller database.
Is this any help?

[tickteng]

I confess I'm not sure I fully understood your question. Do you mean that there are certain hits that are retrieved from one database, and don't appear when searching on the other? If so, then it might be related to E (the Expect value). Since E is the chance of retrieving a similar hit by random, databases of different size (total number of bases) will return hits with different E values: the smaller the database, the smaller the E. So, you might not retrieve hits from the big database if the E is just big enough to be above your cut-off value, while you might still retrieve that hit from a smaller database.
Is this any help?

hi, thanks for the reply. So let me make it more clear. I am using nanopore to sequence an amplicon of pathogen A. I tried to blast the sequence using 2 different databases. The result both showed pathogen A, but in one database there are more reads that hit pathogen A compared to the other database.
If that is the case, is the e-value still applicable? also, how can i tell that e value is big or small? Is e2-10 > e2-20?

[r.rosati]

Hi!
Regarding your question, 2E-10 means 2*10^-10, and 2E-20 means 2*10^-20. So 2E-20 is lower.
However these values are pretty stringent, if you have a 2E-10 hit in one database, it'd take a database several orders of magnitude bigger to fall outside an E<=10 parameter. If these are your E-values, your problem is not the one I mentioned, I reckon.