Poboljšani algoritam za izbor i provjeru kvalitete najboljih multivarijacijskih modela odnosa strukture i svojstava molekula

scientific master's theses

• Title (english): Improved algorithm for selection and validation of best multivariate structure-property molecular models
• Author: Lidija Papeš Šokčević
• Mentor: Mile Šikić (mentor)
• Mentor: Bono Lučić (komentor)
• Committee member: Mladen Vučić (predsjednik povjerenstva)
• Committee member: Bono Lučić (član povjerenstva)
• Committee member: Mile Šikić (član povjerenstva)
• Committee member: Sven Lončarić (član povjerenstva)
• Committee member: Sanja Tomić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; (Department of Electronic Systems and Information Processing); Zagreb
• Granter: Ruđer Bošković Institute; Zagreb
• Defense date and country: 2011-10-26, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing
• Scientific / art field, discipline and subdiscipline: BIOMEDICINE AND HEALTHCARE; Pharmacy
• Scientific / art field, discipline and subdiscipline: NATURAL SCIENCES; Chemistry
• Universal decimal classification (UDC): 004 - Computer science and technology. Computing. Data processing ; 54 - Chemistry. Crystallography. Mineralogy ; 615 - Pharmacology. Therapeutics. Toxicology

Abstract

Algoritam za izbor najboljih multivarijatnih modela (prema koeficijentu korelacije) razvijen je za potrebe primjena u istraživanju lijekova. Realiziran je u programskom jeziku Visual Basic i povezan je s bazom podataka. U radu je istražen odnos između statističkih parametara izračunatih na podacima u postupku prilagodbe na skupu za učenje, križne provjere ispuštanjem određenog postotka podataka skupa za učenje i u postupku predviđanja na vanjskom skupu podataka. Postupak križne provjere u kojem se u svakom koraku izbacuju veći podskupovi podataka pokazuje bolje slaganje s rezultatima dobivenim u predviđanju na vanjskom skupu nego najčešće korišteni postupak križne provjere u kojem se u svakom koraku izbacuje po jedna molekula. Nadalje, slaganje između statističkih parametara izračunatih u postupku prilagodbe na skupu za učenje i odgovarajućih parametara izračunatih u postupcima križnih provjera i na vanjskome skupu, znatno je bolje nego slaganja objavljivana u literaturi. To potvrđuje dobre strane primijenjenog postupka početne eliminacije deskriptora, smanjenja korelacije između deskriptora, postupka modeliranja i primijenjenoga algoritma za izbor modela. Za razliku od uobičajenih postupaka križne provjere koji se rabe u literaturi istražile su se učestalosti pojavljivanja pojedinih deskriptora u najboljim modelima. Na temelju tih analiza načinio se redoslijed važnosti pojedinih deskriptora u najboljim modelima, što je važna informacija pri interpretaciji i uporabi najboljih modela. Dobiveni rezultati i razvijena aplikacija MR_QSAR vrijedan su znanstveni i stručni doprinos području razvoja i primjene računalnih algoritama za modeliranja svojstava i aktivnosti molekula koja se provode u znanstvenim krugovima, u istraživanjima novih lijekova u farmaceutskoj industriji, te u zaštiti okoliša u postupcima procjene toksičnosti molekula.

Abstract (english)

An algorithm for the selection of best multivariate regression models (according to correlation coefficient) was developed for the use in drug research. It is realized in the programming language Visual Basic and connected to a database. This work explores the relationship between the statistical parameters calculated in the process of fitting on the training data set, cross-validation by dropping a certain percentage of data study sets (k-fold cross-validation), and in the process of prediction on an external data set. Cross-validation procedure, in which a larger sub-set of data is omitted in each step, shows better agreement with results obtained in prediction on external set than the leave-one-out cross-validation procedure. Additionally, correlations between statistical parameters calculated in the fitting procedure on the training set and corresponding parameters calculated in the cross-validation procedures and in validations on external data sets are much better than it was published in literature. Such an analysis confirmed very good attributes of applied procedures for initial elimination of descriptors, reduction of inter-correlations between descriptors, and very good attributes of modeling method and algorithm used for model selection. Unlike a regular cross-validation procedures used in the literature, frequencies of occurences of individual descriptors in the best models were explored. Based on this analysis an order of importance of individual descriptors in best models was made. This information is important in the interpretation and use of best models. Obtained results and developed application MR_QSAR are valuable scientific and technical contribution to the development and application of computer algorithms for modeling molecular properties and activities carried out in scientific circles, in studies of new drugs in the pharmaceutical industry, and in environmental protection in procedures for assessment the toxicity of the molecules.

• Keywords: QSAR modeliranje; multivarijacijska linearna regresija; izbor deskriptora; razvoj modela; statistički parametri; prilagodba modela; križna provjera; vanjska provjera; topljivost molekula u vodi
• Keywords (english): QSAR modeling; multivariate linear regresion; selection of descriptors; model development; statistical parameters; model fitting; cross-validation; external validation; water solubility of molecules
• Language: croatian
• URN:NBN: urn:nbn:hr:168:987128
• Project: Number: 036-0362214-1987; Title: Modeliranje kompleksnih sustava; Leader: Branko Jeren; Jurisdiction: Croatia; Funder: MZOS; Funding stream: ZP
• Project: Number: 079-0000000-3211; Title: Odnos strukture i aktivnosti flavonoida; Leader: Dragan Amić; Jurisdiction: Croatia; Funder: MZOS; Funding stream: ZP
• Project: Number: 098-1191344-2860; Title: Proučavanje biomakromolekula računalnim metodama i razvoj novih algoritama; Leader: Sanja Tomić; Jurisdiction: Croatia; Funder: MZOS; Funding stream: ZP
• Project: Number: 098-1770495-2919; Title: Razvoj metoda za modeliranje svojstava bioaktivnih molekula i proteina; Leader: Bono Lučić; Jurisdiction: Croatia; Funder: MZOS; Funding stream: ZP
• Study programme: Title: Postgraduate master programe in computing; Study programme type: university; Study level: postgraduate; Academic / professional title: magistar inženjer računarstva (magistar inženjer računarstva)
• Catalog URL: http://lib.fer.hr/cgi-bin/koha/opac-detail.pl?biblionumber=36980
• Type of resource: Text
• Extent: 71 str. ; 30 cm
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2019-07-02 08:06:11