PATTERNHUNTER VS BLAST
BLAST uses seeded matches of 11 base pairs which cannot recognize continuous matches less than 11 base pairs.
Thus, low sensitivity is the biggest problem of BLAST.
Dilemma of BLAST:
Enlarging the seed size lowers the sensitivity while reducing seed size reduces the speed of calculations.
BLAST may miss a hit !
HOW PATTERNHUNTER WORKS?
-
PatternHunter utilizes numerous seeds (tiny search strings) with optimal intervals between them.
-
Searches that employ seeds are extremely fast because they only determine homology in places where hits are established.
-
The sensitivity of a search string is greatly influeced by the amount of space between adjacent strings.
Why PatternHunter is better?
PatternHunter uses "spaced seeds"
For example:
PatternHunter increases sensitivity and speed simultaneously.
​
For BLAST seed: 11111111111
spaced seed: 111010010100110111,
where 1 means a required match and 0 means the "don't care position".
​
-
Spaced seed perform non-consecutive matches and optimized match positions.
-
This significantly increases hit to homologous region while reducing bad hits.
PatternHunter exhibits speeds that appoximately a hundred times faster than BLAST and MegaBLAST.
APPLICATION OF PATTERNHUNTER
-
PatternHunter has been used to compare the human genome with 16 million reads of the unassembled mouse genome, a total of nine billion base pairs.
-
Homology search is a very lengthy procedure that requires a lot of time.
-
Handling DNA-DNA searches as well as translated DNA-protein searches are still challenging due to the vast sizes of databases and the tiny query that is used.
-
PatternHunter has been improved to an upgraded version (PatternHunter II) which hastens the DNA-protein searches a hundredfold without altering the sensitivity.
-
PatternHunter identifies all the homologies between large DNA sequences and it is reported as being much faster and higher in quality than BLAST.
-
-
While MegaBlast is designed for high speed on highly similar sequences and Blastn for sensitivity, PatternHunter simultaneously exceeds Blastn in sensitivity, MegaBlast in speed (on long sequences), and both in memory use.