Local BLAST no hits found

[PSchneeb]

Hi all,

I am trying to BLAST several short queries (20nt) against a local database.
The local database is a merged file containing all genomes which I used to design the queries.
The problem is that for some queries I get "no hits found". This should not happen since I created the queries from the database and I think I should at least get "1 hits found" for each query...

I am using the latest version of BLAST (2.2.26+) and the following command :

blastn.exe, -task, blastn-short, -db, database.fas, -perc_identity, 100, -ungapped, -num_threads, 4, -query, queryfile.fas, -out, outputfile.txt, -outfmt, 7

Any help would be much appreciated!!

[Bastian]

have you tried without "-short"

[PSchneeb]

Hi Bastian,

Yes I tried without the -short command.
But for both (short and without short), I still get the same results where some of the queries have "0 hits found" and some other have "1 or more hits found".

[Bastian]

is it possible, the are just not there?

You are running this in windows, or on what system? how does the outputfile look like?

[PSchneeb]

I looked in the fasta files I used to create the database and they were present, even several times for some of them...

I am using window 7 64bits (BLAST is also in 64bits).

And the output looks like this :

# BLASTN 2.2.26+
# Query: F.Flaviviridae_G.Flavivirus_S.Dengue_virus_GI.JN638336_Pri.TGTTGTCAGTGTGGAATAGG_Num.3
# Database: T:/LastVersion/blast-2.2.26+/data/VirusesBlast_Family/F.Flaviviridae.fas
# Fields: query id, subject id, % identity, alignment length, mismatches, gap opens, q. start, q. end, s. start, s. end, evalue, bit score
# 1 hits found
F.Flaviviridae_G.Flavivirus_S.Dengue_virus_GI.JN638336_Pri.TGTTGTCAGTGTGGAATAGG_Num.3 F.Flaviviridae_G.Flavivirus_S.Dengue_virus_GI.JN697056 100.00 20 0 0 1 20 9996 10015 3e-005 37.9
# BLASTN 2.2.26+
# Query: F.Flaviviridae_G.Flavivirus_S.Dengue_virus_GI.JN638336_Pri.ATGGGATACAAGGATAACAG_Num.4
# Database: T:/LastVersion/blast-2.2.26+/data/VirusesBlast_Family/F.Flaviviridae.fas
# 0 hits found
# BLASTN 2.2.26+
# Query: F.Flaviviridae_G.Flavivirus_S.Dengue_virus_GI.JN638336_Pri.GATTTGTGTGAGGGTACCAC_Num.5
# Database: T:/LastVersion/blast-2.2.26+/data/VirusesBlast_Family/F.Flaviviridae.fas
# Fields: query id, subject id, % identity, alignment length, mismatches, gap opens, q. start, q. end, s. start, s. end, evalue, bit score
# 29 hits found
F.Flaviviridae_G.Flavivirus_S.Dengue_virus_GI.JN638336_Pri.GATTTGTGTGAGGGTACCAC_Num.5 F.Flaviviridae_G.Flavivirus_S.Dengue_virus_GI.JN819423 100.00 20 0 0 1 20 3229 3248 3e-005 37.9

As you can see, the 1st and the 3rd probes have hits but not the second one. This is weird because as I said, the database contains the genome on which the probe was designed :-/

This has probably something to do with evalue (although this is strange because there should be only "exact matches") but I don't know how to shut this filter off..

[Bastian]

this works in the unix-way easily:
blastall -p blastn -a 4 -m 8 -i Reads.fa -d Database.fa -o Outputfile.out

...also for short reads

so try just the very default option and remove the rest. BTW m is the outputformat and "8" is much more usefull, i think

[PSchneeb]

Thanks for the suggestion!
I tried without the "ungapped" command and now I have at least one 1 per probe which is fine.
This is weird because I thought that the "ungapped" command is only used to discard hits with gaps but I'm probably wrong.

The output format "8" is ASN.1 in the latest version and the files created are just huge!
By the way, is it possible to print just the following part in the BLAST output file?

# BLASTN 2.2.26+
# Query: F.Flaviviridae_G.Flavivirus_S.Dengue_virus_GI.JN638336_Pri.GATTTGTGTGAGGGTACCAC_Num.5
# Database: T:/LastVersion/blast-2.2.26+/data/VirusesBlast_Family/F.Flaviviridae.fas
# Fields: query id, subject id, % identity, alignment length, mismatches, gap opens, q. start, q. end, s. start, s. end, evalue, bit score
# 29 hits found

I know that it is possible to customize the outfmt command but I didn't manage to get it to work...