Efektivni algoritmy pro vyhledavani shodnych molekul a jejich aplikace

Bc. Ján Slaninka

Diplomová práce

Effective algorithms for searching of identical molecules and their application

Abstract:

The topic of this thesis is focused on identical molecules searching based on the isomorphism of molecular graphs. As graph isomorphism searching problem is computationally very difficult task, and belongs to NP class problems, an efficient way of how to deal with the search should be implemented. A multiple graph isomorphism searching algorithms are presented, along with other approaches used for molecule matching, such as dividing atoms into equivalence classes or chemical properties of a molecule. The most common chemical data formats are described, focusing on the SDFile data format and SMILES string representation as they are used in the developed application. Based on the fact, that we are dealing with a special representation of a graph, a molecular graph, the most suitable approach was chosen. It uses optimized version of the backtracking algorithm with atoms distribution into individual equivalence classes, chemical properties of the molecule to search only a subset of the database and various other optimizations. Selected approach was implemented and tested on the real-world chemical databases. Developed application will be used as a tool for molecule searching within bioinformatic project of the ANF Data company.

Keywords

molecule graph, isomorphism, backtracking, canonicalization, SMILES, SDF, Java

Jazyk práce: angličtina

Datum vytvoření / odevzdání či podání práce: 23. 5. 2008

Obhajoba závěrečné práce

• Obhajoba proběhla 2. 7. 2008
• Vedoucí: doc. RNDr. Radka Svobodová Vařeková, Ph.D.
• Oponent: prof. Ing. Václav Přenosil, CSc.

Masarykova univerzita

Fakulta informatiky

Magisterský studijní program / obor:
Aplikovaná informatika / Aplikovaná informatika