Problem when running remote Blast with a big file

Question

[arturo.marin]

Hi all, I am new in this forum, so I hope to explain myself well. I am trying to do remote Blast with an input file with 237,805 sequences. I used the next command:

blastn -db nt -query JAL6513ln1.fa -out JAL6513ln1.xml -outfmt 5 -max_target_seqs 10 -remote

But at some point of the execution this give me several error:

Error: (308.5) [blastn] [blast4]  Service not found
Error: (301.5) [blastn] [CONN_Open(blast4)]  Failed to open connection: Closed
Error: (315.4) [blastn] CConn_Streambuf::overflow():  CONN_Write() failed (blast4): Closed
Error: (315.4) [blastn] CConn_Streambuf::overflow():  CONN_Write() failed (blast4): Closed
Error: (315.4) [blastn] CConn_Streambuf::overflow():  CONN_Write() failed (blast4): Closed
Error: (802.5) [blastn] CObjectOStream: error at byte 52: stack is empty: cannot close output stream
Error: (CIOException::eFlush) [blastn] COStreamBuffer::Flush: failed
Error: (802.4) [blastn] Cannot close serializing output stream (CSerialException::eIoError) byte 52: cannot close output stream

When I try remote Blast with a subset of the previous file with only 100 sequences it since there is no problem. Somebody have any idea of this errors? is possible that I need to "cut" the original file in several more tinny files?

Answer 1

[gb]

Remote blast is not made for so many sequences at once. There is a certain limit and the ncbi will refuse or close the connection if you exceed it. There is a page somewhere that explains it but can't find it right now. I believe they first slow down your blast and after that they close the connection. There is or was also a way to contact the ncbi to temporary white list your ip.

EDIT: Think this is the page that I mentioned
https://blast.ncbi.nlm.nih.gov/Blast.cgi?CMD=Web&PAGE_TYPE=BlastDocs&DOC_TYPE=DeveloperInfo

The best solution is to run blast locally, an other not so recommended solution is to build in breaks of a few seconds between each sequence and if you get an error wait 5 minutes and start it again with the sequence that gave the error. But again that is not so recommended.

[gb]

Did this answer helped you, or do you need more information?

[arturo.marin]

Yes, thanks for your help. I think I'll try to use Blastn localy.