Weird Tabular BLAST Occurrence

[tboothby]

Something odd I just found while performing a stand alone BLAST. I have the number of descriptions set to 10, which is the number of hits reported for most of my queries... but one of my queries has over 50 hits reported.

Anyone have any idea why this would be? I find it bizarre, I have never seen this before.

[maubp]

Which version of BLAST, and what was the command line you used? i.e. Which options?

[tboothby]

I am using the latest release of BLAST+

Command line:
blastx -db databases/refseq_protein -gilist GI_hyp_put_pre_unchar -query /Users/wolniaklab/Desktop/1-2vs6-8_1_2_Sig_Increase/1-2vs6-8_1_2_Sig_Increase.txt -out /Users/wolniaklab/Desktop/1-2vs6-8_1_2_Sig_Increase/1-2vs6-8_1_2_Sig_Increase_OUT_TAB -outfmt 7 -num_descriptions 10 -num_alignments 10 -num_threads 8

[maubp]

There was a bug in -num_descriptions and/or -num_alignments IIRC, and the NCBI recommended using -max_target_seqs which should override them.

[aloliveira]

Looks like -max_target_seqs does not work neither. Another solutions?

[maubp]

Do you have an example? Perhaps you are counting HSPs while BLAST is counting sequence matches?