Problem with NCBI-BlastN

Question

[biofalconch]

Hello, I have recently reinstalled the OS of my workstation, changing from Fedora to Ubuntu. I then proceeded to install ncbi-blast (in its 2.2.28 version), but oddly when I try to run a job, it only gives me a few hits in the output file, and continues running without giving any more hits (and no, it is not because there are no hits with the rest of the sequences). I have checked my database and input file and they seem fine, any idea of what the problem might be?

[5heikki]

Commands you used? Did you let it run to the end?

[biofalconch]

-task blastn \
-evalue 10 \
-word_size 4 \
-gapopen 10 \
-gapextend 6 \
-penalty -4 \
-reward 5 \
-max_target_seqs 1 \
-outfmt 6 \
-num_threads 10 &

These are the parameters (i did include input,output and db), and I ran in for 20 minutes, didn't change anything on the output in that time

[5heikki]

-max_target_seqs 1

What do you think this does? This is why you get only hits to one target sequence. Also, any particular reason why you're using legacy blast instead of blast+?

[biofalconch]

Im using blast+, and also -max_target_seqs gives me one hit per query. I dont get only one hit, i get the first 100+ querys aproximately, but then it stops.

[5heikki]

Ok, so your input contains multiple seqs. Maybe check what's wrong with the query seq from which on you get no hits?

[biofalconch]

It seems normal, also tried with different inputs, getting the same problem

[5heikki]

But maybe it's not normal? Maybe e.g. some header contains a weird character like † (I've seen this in some fasta files)

You could try:

iconv -f utf-8 -t ascii yourInput.fasta

If it says "cannot convert", then:

iconv -c -f utf-8 -t ascii yourInput.fasta

[biofalconch]

I ran the first command and it finished without errors. Any chance of it being the parameters?

[5heikki]

I don't know. All I can do is guess. Is your db custom? Perhaps the problem lies there. Does simple blastn -query yourseqs -db yourdb execute fine? Are you sure you have enough cores for 10 threads? Are you sure you have enough RAM for your db?

[biofalconch]

Sorry for taking to long to answer, I didn't have the means to run the test you suggested until now. Well, I guess I will start addresing the questions in order. Yes, my db is custom, but I did check it and it seems fine. Now, I ran a few trials without a few parameters and found that the ones giving me trouble were all the scores, but I can't understand why. Also word_size is giving me a few problems. And yes, I have enough computing power to run it on those settings.