I'm blasting (blastx2.2.28) my assembly to the annotated refseq. Below - the command line I used:
blastx -query assembly.fa -out Blast -db BlastCustomDB -num_threads 8 -evalue 1e-6 -best_hit_score_edge 0.05 -best_hit_overhang 0.25 -max_hsps_per_subject 1 -max_target_seqs 1 -outfmt 6
I specified quantity of HighcoringSegmentPairs (1), but still BLAST output contains two or more HSP for some queries.
Here's the output example (it goes like query ID, subject ID, identity %, length of HSP, mismatches, gap opens ... E-value, bit score):
TR60-c0_g1_i1 EKV52009.1 37.36 174 57 7 ... 2e-14 74.3 TR60-c0_g1_i1 EKV52009.1 64.37 87 30 1 ... 6e-10 59.7 TR61-c0_g1_i1 EKV51611.1 49.41 253 121 2 ... 4e-47 161 TR64-c0_g1_i1 EKV51289.1 60.00 175 62 1 ... 9e-66 208
After examination with 'uniq' command it became clear that all duplicated results for a query each time were aligned to a single gene in different position (what is exactly HPSs).
There's the way just to remove multiple HSP, but maybe it's possible to change some parameters in blast run to get the right single-hit-single-HSP output directly?
Thanks in advance!