NCBI Blast Search Program Alignment Issue.

Question

[clear.choi]

Hi,

I am using NCBI Alignment program in order to align sequence.

And I have a question during alignment.

There is Output sequence from the machine.

ATTGAAAAGGAGGGAGCTACTCTCAGGCTGCAA

And also there is reference Sequence.

TTAGAAAAGGAGGGAGCTACTCTCAGGCTGCAA

After Alignment I got result as below

Query  4     GAAAAGGAGGGAGCTACTCTCAGGCTGCAA  33
             ||||||||||||||||||||||||||||||
Sbjct  4     GAAAAGGAGGGAGCTACTCTCAGGCTGCAA  33

But Actually I expect there is 3 deletion in front of that sequence.

So I want like below with 3 mismatch.

Query  1     ATT-GAAAAGGAGGGAGCTACTCTCAGGCTGCAA  33
              || ||||||||||||||||||||||||||||||
Sbjct  1     -TTAGAAAAGGAGGGAGCTACTCTCAGGCTGCAA  33

or

Query  1     ATTGAAAAGGAGGGAGCTACTCTCAGGCTGCAA  33
              | ||||||||||||||||||||||||||||||
Sbjct  1     TTAGAAAAGGAGGGAGCTACTCTCAGGCTGCAA  33

Do you guys have any idea I give this option to blast search program?

My Script as below

./blastn -db referecen.db -query fullsequence.fasta -out output.out -task blastn

Web site for compare at

http://blast.ncbi.nlm.nih.gov/Blast.cgi?PAGE_TYPE=BlastSearch&PROG_DEF=blastn&BLAST_PROG_DEF=megaBlast&BLAST_SPEC=blast2seq

Thank you!

Answer 1

[Istvan Albert]

BLAST does local alignments.

What you seem to be after is the so called semi-global (end-to-end) alignment. Just make sure that is indeed what you are looking for. If so you will need to use a different tool for that. Many short read aligners will work.

[clear.choi]

I am sorry I have updated even thought another alignment tool at least has to show as below

Query  264   ATT-GAAAAGGAGGGAGCTACTCTCAGGCTGCAA  2677
              || ||||||||||||||||||||||||||||||
Sbjct  1     -TTAGAAAAGGAGGGAGCTACTCTCAGGCTGCAA  33

right?

[Istvan Albert]

The same answer stands - if you want the entire query to match you need to use an aligner that is designed to do so. BLAST looks for the strongest local alignment. It does not want to reward a longer alignment unless the score gets better. Adding mismatches to the end of the alignment will always reduce the score since there are no positive contributions later on.

Make sure to understand that "alignment" is just an abstract word - there is no "correct" alignment - instead there are various algorithms and methods that, together with the score will produce "one aligment" that is optimized for those. If you need semi global alignments make sure to use a semi global aligner.

[clear.choi]

Thank you so much for your replying.. But I guess there might be an option to do in the program. because If I use my own smith-waterman algorithm it will be expected result... I am not really familiar with Alignment process yet I will try it more thank you again :)

[clear.choi]

Hi Now I got your point getting close with Alignment :) Thank you so much!