Biology Reference
In-Depth Information
process, the given sequence group A is randomly separated into two
groups, and further profile-profile alignment is performed on the
two groups. In our program, the refinement iteration time is set to
be 10 by default, or can also be set by users.
The final progressive alignment is performed along the guide
tree from closely related to distantly related. Then a final iterative
alignment is applied to refine the results from progressive align-
ment. In addition, a multi-threading technology based on
OPENMP is also used to improve the time efficiency of the pro-
gram [ 19 ].
4 Case Study
We used BAliBASE 3.0 [ 20 ] data sets as training sets, and SABmark
1.65 [ 21 ] and OXBENCH [ 22 ] data sets as testing sets, for the
parameter optimization. The alignment results were evaluated
according to sum-of-pairs (SP) score and true column (TC) scores
[ 4 , 23 , 24 ]. We used the application bali_score provided by BAli-
BASE 3.0 to calculate these scores. Users can download the protein
sequence alignment benchmark datasets from http://www.drive5.
com/bench/ .
Fig. 2 shows an example input multiple sequence file BB12003
from BAliBASE 3.0 database and its output by running
Fig. 2 The example input file BB12003 and its output generated by MSACompro
Search WWH ::




Custom Search