TY - JOUR
T1 - Algorithm engineering for optimal alignment of protein structure distance matrices
AU - Wohlers, Inken
AU - Andonov, Rumen
AU - Klau, Gunnar W.
N1 - Funding Information:
Acknowledgments This work has been supported by DFG grant KL 1390/2–1 and was partly done when I. Wohlers was visiting IRISA supported by an INRIA grant. R. Andonov is supported by BioWIC ANR-08-SEGI-005 project and partially by DVU-01/197/16.12.2008 NSF-Bulgaria. Computational experiments were sponsored by the NCF for the use of supercomputer facilities, with financial support from NWO.
PY - 2011/8
Y1 - 2011/8
N2 - Protein structural alignment is an important problem in computational biology. In this paper, we present first successes on provably optimal pairwise alignment of protein inter-residue distance matrices, using the popular dali scoring function. We introduce the structural alignment problem formally, which enables us to express a variety of scoring functions used in previous work as special cases in a unified framework. Further, we propose the first mathematical model for computing optimal structural alignments based on dense inter-residue distance matrices. We therefore reformulate the problem as a special graph problem and give a tight integer linear programming model. We then present algorithm engineering techniques to handle the huge integer linear programs of real-life distance matrix alignment problems. Applying these techniques, we can compute provably optimal dali alignments for the very first time.
AB - Protein structural alignment is an important problem in computational biology. In this paper, we present first successes on provably optimal pairwise alignment of protein inter-residue distance matrices, using the popular dali scoring function. We introduce the structural alignment problem formally, which enables us to express a variety of scoring functions used in previous work as special cases in a unified framework. Further, we propose the first mathematical model for computing optimal structural alignments based on dense inter-residue distance matrices. We therefore reformulate the problem as a special graph problem and give a tight integer linear programming model. We then present algorithm engineering techniques to handle the huge integer linear programs of real-life distance matrix alignment problems. Applying these techniques, we can compute provably optimal dali alignments for the very first time.
UR - http://www.scopus.com/inward/record.url?scp=79960116362&partnerID=8YFLogxK
U2 - 10.1007/s11590-011-0313-3
DO - 10.1007/s11590-011-0313-3
M3 - Journal articles
AN - SCOPUS:79960116362
SN - 1862-4472
VL - 5
SP - 421
EP - 433
JO - Optimization Letters
JF - Optimization Letters
IS - 3
ER -