Due to the specific BLAST algorithm, short sequences cannot be used as queries. Thus, I wonder if there are other methods to search homology of short sequences with ESTs or mRNAs. For example, can BLAST performed after the sequences are assembled in contig?
if you reduce the wordsize blast should do the job. Wordsize is the length of the perfect match that is used as a seed for the alignment.
Assembly of that much short sequences is a harder process and probably it will cause new reliability problems. BLAST should be used for short queries. Actually, It automatically changes algorithm according to short queries.
My sRNA sequences (queries) are 18-26nt long. Blastx, for example, can run with word size vaules of 2 or 3. Do you think that a word size of 2 is proper? Of course, my question is: are results of a blastx search trustworthy when short sequences are used as queries?
Your statement that short sequences cannot be used is not correct. For some time now, the BLAST algorithm has included specific modifications just for short sequence searches (since around 2007 or 2008). It has the ability to "Search for short, nearly exact sequences" which was designed to work with sequences of between 7 and 20 bp in length (e.g. motif finding).
The equivalent protein search is designed around queries of 5-15 residues.
There are details of all of the various program choice options on the BLAST help pages under the Program Selection Guide. And if you want detailed techical help with optimizing your search, I would email your question to the NCBI staff at [email protected] as these are the foremost experts on just what BLAST is capable of.
There are some programs that perform more fast and accurate aligment for short read sequences. If you want to align your sequences to nucleotides databases (like BLASTn) you can use Bowtie (http://bowtie-bio.sourceforge.net/index.shtml) or SOAP (http://soap.genomics.org.cn/). If you want to align to protein databases (like Blastx) you can use RAPSearch (http://omics.informatics.indiana.edu/mg/RAPSearch/).
I think that these sequences can be usefully explored only if we use the software, cyber and information systems and laws. In this sense, we need to determine the sequence of numeric value, and then perform a mathematical analysis of the obtained results.
- Badges
- Science method