How to tune blast parameters when looking for miRNAs?

Question

[seta]

Hi everybody,

The E-value cutoff in the miRBase is 10 by default. In my opinion,it seems a bit high for miRNA (short sequence) since the probability of having good alignment by chance is increased as the sequence length is decreased. If you agree with me, could you please let me know on the basis of E-value, which hits resulted from blast output should be considered for conducting further analysis?

In general, since we may get many alignments with e-value of 0 but different scores, it seems that the sorting should be based on score not E-value, isn't it? however, I am not sure about this and glad to hear clear your opinion. Any idea is warmly welcomed.

[margxenscienculo]

Hi . I have similar problem. I don't know if it work but I read some papers that they do a blast with 3 ó 2 mismatches and 80 % of coverage. So I m going to look for the blast options.

[margxenscienculo]

Take a look here!

Answer 1

[Siva]

You can try the the -task option and set it to blastn-short which is optimized for shorter query sequences (<30 nucleotides). From an earlier comment: A: blastn for short sequences

You can also try blat. Using Blat for short sequences with maximum sensitivity

Answer 2

[margxenscienculo]

I have similar problem. The results only have 11 pb of query which has blast. I wonder ... If I want to find new microRNAs of my hypotheticals. Firstly, I have to do a blast against the last version of miRBase. The blast must be completely all the microRNA length. Not only 11 pb. Does not it?