Blastn frozen

Question

[Begonia_pavonina]

I am just starting to use blastn with the -remote option. I have written the following script based on the NCBI documentation.

blastn \
-db nt \
-query queries.fsa \
-out results.out \
-remote

The queries.fsa file contain a single sequence of nucleotide.

But the job launched from this script is frozen, and the only output is the content of the result.out file: " BLASTN 2.15.0+". I think I might forget something very simple that prevent this script to work. Would anyone have an idea what it is?

[GenoMax]

But the job launched from this script is frozen

What does this mean? It can take some time to get analysis back especially depending on load on NCBI blast servers. If you don't change default parameters then you are likely to get hundreds of hits back (you may want that but important to consider). Also define a -task option depending on your query sequence. e.g. if query is short. If you wish to parse the output programmatically then using -outfmt 6 is useful. If you simply want a pretty formatted page then use -html.

[Istvan Albert]

Try your search with a single query that you know should work in a reasonably short amount of time.

With that, you can figure out whether you are running it correctly or not.

Also, in my experience, the command line blast returns answers a lot slower than the web-based blast. So factor that in as well.

[Begonia_pavonina]

Thank you for the information, I have modified my query according to your advices, using as input a short sequence of 150bp instead of previous long sequence and limiting the hits back. And as you mentioned, command line blast a lot slower than online query. A last thing is that blast -remote was impossible through the firewall of my home institution, and I had to make sure to submit the job either on another place or in a server.

Answer 1

[Mensur Dlakic]

We don't know the length of your sequence. If this is a genome-size sequence, or something longer than say 10,000 nucleotides, it may take a while.

Adding the -evalue option, say -evalue 1e-20, will reduce the output size, which will also speed things up. Sometimes it is not the search that takes the most time, but rather making all the pairwise alignments or otherwise processing the output.

Like others suggested, try this on a single sequence that is not too long to get a feel for how long it takes. I concur that this usually takes longer than a simple web search.

[Begonia_pavonina]

Thanks a lot for this advice, -evalue indeed reduce drastically the processing time. The size of the query is as well important on computational time.