Convert Blastall cmd to Blast+

Question

[cowmoomoo]

Hi,

I'm trying to convert an old blastall script (the older version of Blast that NCBI recommends people to upgrade from) to a blast+ command.

I've tried using the NCBI's legacy_blast.pl command to generate the new blastall script, but I'm still getting different results.

#legacy blastall
./legacy_blast.pl blastall -i query.fasta -d /path/to/blastdb -o blastResult.blast -m 7 -e 1.00000e-10 -b 5 -v 5 -p blastx -Q 11 -S 1 -F "M s" --print_only

#converted blast+ cmd
/usr/bin/blastx -db /path/to/blastdb -query query.fasta -evalue 1.00000e-10 -num_descriptions 5 -num_alignments 5 -query_gencode 11 -out blastResult.blast-outfmt 5 -seg yes -soft_masking true

It would seem like all the parameters line up, however when I look in the Blast output file, I see the results to be slightly different and also the filter parameters might be different, the blastall has only F as parameter filter while blast+ has L;m for parameter filters,

#legacy blastall blast result xml output

    <Parameters>
      <Parameters_matrix>BLOSUM62</Parameters_matrix>
      <Parameters_expect>1e-10</Parameters_expect>
      <Parameters_gap-open>11</Parameters_gap-open>
      <Parameters_gap-extend>1</Parameters_gap-extend>
      <Parameters_filter>F</Parameters_filter>
    </Parameters>
  </BlastOutput_param>
<BlastOutput_iterations>
<Iteration>
...

vs:

#legacy blast+ result xml output

    <Parameters>
      <Parameters_matrix>BLOSUM62</Parameters_matrix>
      <Parameters_expect>1e-10</Parameters_expect>
      <Parameters_gap-open>11</Parameters_gap-open>
      <Parameters_gap-extend>1</Parameters_gap-extend>
      <Parameters_filter>L;m;</Parameters_filter>
    </Parameters>
  </BlastOutput_param>
<BlastOutput_iterations>
<Iteration>
...

Anyone have an idea what I"m missing and what other flags I need to add to the blast+ command to get it to get the same results as the original blastall cmd?

Thanks in advance for your time and help!

Answer 1

[5heikki]

This script converts and runs the equivalent NCBI C toolkit command line BLAST program and arguments provided to it (whenever possible) to NCBI C++ tookit BLAST programs. Note that to specify options to this script they MUST use 2 dashes to prefix the options AND be listed at the end of the command line invocation to convert.

Answer 2

[h.mon]

Although I am not sure, I guess it is not possible to guarantee identical outputs from BLAST+ and legacy BLAST - many things changed along the way, not all of them in a backward-compatible manner. I remember reading about issues similar to yours on the blog Blasted Bioinformatics.