A comprehensive comparison of multiple sequence alignment programs
Julie Thompson, Frédéric Plewniak, Olivier Poch
Abstract
Open-access reader
Julie Thompson, Frédéric Plewniak, Olivier Poch
Abstract
Open-access reader
In recent years improvements to existing programs and the introduction of new iterative algorithms have changed the state-of-the-art in protein sequence alignment. This paper presents the first systematic study of the most commonly used alignment programs using BAliBASE benchmark alignments as test cases. Even below the 'twilight zone' at 10-20% residue identity, the best programs were capable of correctly aligning on average 47% of the residues. We show that iterative algorithms often offer improved alignment accuracy though at the expense of computation time. A notable exception was the effect of introducing a single divergent sequence into a set of closely related sequences, causing the iteration to diverge away from the best alignment. Global alignment programs generally performed better than local methods, except in the presence of large N/C-terminal extensions and internal insertions. In these cases, a local algorithm was more successful in identifying the most conserved motifs. This study enables us to propose appropriate alignment strategies, depending on the nature of a particular set of sequences. The employment of more than one program based on different alignment techniques should significantly improve the quality of automatic protein sequence alignment methods. The results also indicate guidelines for improvement of alignment algorithms.
OpenAlex reports 754 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
In recent years improvements to existing programs and the introduction of new iterative algorithms have changed the state-of-the-art in protein sequence alignment. This paper presents the first systematic study of the most commonly used alignment programs using BAliBASE benchmark alignments as test cases. Even below the 'twilight zone' at 10-20% residue identity, the best programs were capable of correctly aligning on average 47% of the residues. We show that iterative algorithms often offer improved alignment accuracy though at the expense of computation time. A notable exception was the effect of introducing a single divergent sequence into a set of closely related sequences, causing the iteration to diverge away from the best alignment. Global alignment programs generally performed better than local methods, except in the presence of large N/C-terminal extensions and internal insertions. In these cases, a local algorithm was more successful in identifying the most conserved motifs. This study enables us to propose appropriate alignment strategies, depending on the nature of a particular set of sequences. The employment of more than one program based on different alignment techniques should significantly improve the quality of automatic protein sequence alignment methods. The results also indicate guidelines for improvement of alignment algorithms.
Key concepts: Alignment-free sequence analysis, Multiple sequence alignment, Structural alignment, Sequence alignment, Benchmark (surveying), Set (abstract data type), Biology, Sequence (biology)