DbClustal: rapid and reliable global multiple alignments of protein sequences detected by database searches

Abstract

DbClustal addresses the important problem of the automatic multiple alignment of the top scoring full-length sequences detected by a database homology search. By combining the advantages of both local and global alignment algorithms into a single system, DbClustal is able to provide accurate global alignments of highly divergent, complex sequence sets. Local alignment information is incorporated into a ClustalW global alignment in the form of a list of anchor points between pairs of sequences. The method is demonstrated using anchors supplied by the Blast post-processing program, Ballast. The rapidity and reliability of DbClustal have been demonstrated using the recently annotated Pyrococcus abyssi proteome where the number of alignments with totally misaligned sequences was reduced from 20% to <2%. A web site has been implemented proposing BlastP database searches with automatic alignment of the top hits by DbClustal.

Figure 1

Propagation of the anchors between the query and each database sequence (shaded regions). The overlapping segments in the database sequences are used as an anchor between the two sequences (shown in black).

Figure 2

Alignment of a set of six proteins by (A) a query-based alignment method and (B) DbClustal using a progressive tree-based method. The anchors supplied by Ballast are shown in red. The boxed segment in P29397 indicates an incorrect anchor.

Figure 3

A plot of the ratio of the MD scores for 1683 alignments obtained by ClustalW and DbClustal. The red lines denote the score ratio threshold of 1.05, above which the two alignments are considered to be significantly different.

Figure 4

Alignment of a set of 11 proteins by (A) ClustalW and (B) DbClustal. The N- and C-terminal extensions in both alignments are not shown. The corresponding segments in the ClustalW alignments are shown for comparison purposes. The regions denoted by the blue arrows below the DbClustal alignment correspond to the blocks detected by the Probe program. Ten homologues were detected by Probe, with four sequences being included in the final model alignment. These are indicated by the underlined names.

Figure 5

Alignment of a set of 15 proteins by (A) ClustalW and (B) DbClustal. The N- and C-terminal extensions in both alignments are not shown. Anchors are shown in red in the DbClustal alignment. The corresponding segments in the ClustalW alignments are shown for comparison purposes. The regions denoted by the blue arrows below the DbClustal alignment correspond to the blocks detected by the Probe program. Of 131 homologues detected by Probe, 46 sequences were included in the final model alignment. Those that were also included in the DbClustal alignment are indicated by the underlined names.