Biology Reference
In-Depth Information
2 Methods
PROMALS3D is a progressive multiple protein sequence alignment
tool. A key feature of PROMALS3D is that it uses different strate-
gies to align subsets of sequences with different levels of difficulty to
properly balance alignment accuracy and speed. Relatively similar
sequences are aligned by a fast algorithm to form pre-aligned
groups without retrieving additional information from databases.
To align the relatively divergent pre-aligned groups, PROMALS3D
resorts to more elaborate alignment techniques and uses additional
information from homologous sequences and structures found by
database searches. Such a method of using different strategies at
different aligning stages allows PROMALS3D to align thousands
of protein sequences in manageable time, since the time-
consuming and memory-consuming steps of database search and
consistency computation are only applied to a subset consisting of
representative sequences of the pre-aligned groups instead of the
entire set of input sequences. The flowchart of PROMALS align-
ment procedure is shown in Fig.
1
.
N
0
input protein sequences
Cluster and align highly similar sequences (>=95%)
N
1
target sequences
Divide into groups and align within each group
N
2
pre-aligned groups
Select representative sequences
N
2
representative sequences
PSI-BLAST against sequence database
Sequence profiles
of
representatives
PSIPRED
PSI-BLAST against structure database
Homologs with
3D structures
Predicted secondary
structures
Sequence-structure alignments
Struct
ure-structure alignments
Profile-profile
comparison
Profile-derived
alignment constraints
Structure-derived
alignment constraints
User-defined
alignment constraints
Consistency-based progressive alignment
Multiple alignment of N
2
representatives
Merge pre-aligned groups and
add highly similar sequences
Multiple alignment of N
0
input sequences
Fig. 1 Flowchart of the PROMALS3D method
