HSP of length 0
Entering edit mode
15 months ago


I have some questions about blast-2.10.1+. I have made a db as follows:

"path/to/NCBI/blast-2.10.1+/bin/makeblastdb.exe" -dbtype "prot" -in "path/to/fasta_db.txt" -parse_seqids -hash_index

I have then used this in a blast+ task run using python's subprocess module - the equivalent command line entry is:

blastp -out results\out\path -query fasta\query.txt -db db_name -evalue eval -outfmt 5 -num_threads thread_num -parse_deflines -seg 'yes' -soft_masking 'true'  -use_sw_tback -num_alignments '200'

When I look at the xml results, I get inconsistent hit id's (e.g. dbj|expected_accession|, gb|expected_accession|, ref|expected_accession|). Can you tell me why this is happening, how to get the accession as the hit id with no prefixes, and, if that is not possible, whether I can assume the hit_id will always be contained within two '|' characters?

As you can see for the HSP below (sorry the formatting wont work properly), I also seem to be getting HSPs of length 0, which I assume is an issue relating to my blast command - however the other metrics (score etc) are there. Would you be able to tell me why this is based on the commands I have provided? I think it is an issue specific to this database (I have performed the same blast call with a different db).


Thanks, Tim

EDIT I noticed that there was an extra file when I made the DB (.pdb-lock), compared to when I had done this previously with success. I tried remaking the database as described above after deleteing all of the old files, and the same thing happened. I then tried making the database without -parse_seqids, the lock file wasn't generated, and I got normal HSPs with a len >0. All of this is leading me to suspect its an issue with how my database is being made, but I've made other dbs with -parse_seqids with no issues before...

alignment blastp • 549 views
Entering edit mode

This may have something to do with it:


Entering edit mode

Hi Genomax, I made a webscraper to get the sequences, although they were removed even when I made the webscraper. The sequences were then made into a local database, so i dont think it should be an issue (it might not be in ncbi's databases but it is in mine). I've also checked all the formating of deflines etc, so it's a bit of a mystery!

From the fasta I used to make the db:

Entering edit mode

Since we can't see/access your data this may be difficult to diagnose. Someone else may have additional insight. Use blastdbcheck -db DB to verify that database looks good.


Login before adding your answer.

Traffic: 1227 users visited in the last hour
Help About
Access RSS

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6