E-Locus - Institutional Repository of the University of Crete

Home Collections School/Department School of Sciences and Engineering Department of Biology Post-graduate theses

Post-graduate theses

Search command : Author="Παυλίδης" And Author="Μιχάλης"

Current Record: 49 of 800

[Add to Basket]

Identifier

000452174

Title

Domain based prediction of human Protein-Protein Interactions

Alternative Title

Πρόβλεψη αλληλεπιδράσεων των πρωτεϊνών στον άνθρωπο, βάσει των αυτοτελών δομικών τους περιοχών

Author

Τσαγλιώτη, Ηλέκτρα Γ.

Thesis advisor

Παυλίδης, Παύλος

Reviewer

Παυλίδης, Ιωάννης
Ηλιόπουλος, Ιωάννης

Abstract

Proteins are the cornerstones of cell function, being key players in all processes that take place within cells. Knowing which proteins interact, is valuable not only for achieving a deeper understanding of proteins themselves, but also for understanding the inner workings of living organisms, such as, ourselves. Knowledge of Protein Protein Interactions (PPIs), contributes to ’basic research’ development and applied science as well, since PPIs can also be linked to therapeutic targets and drug design. In this study, the aim was to create and evaluate a Machine Learning algorithm for PPI prediction in Homo sapiens, solely based on the domains of protein pairs. Our Machine Learning approach aims to classify query protein pairs as ’interacting’ or ’non-interacting’ according to their domain composition. For the training of our ML model, a positive dataset was created by extracting PPI information from the STRING database and domain information from Pfam. A negative dataset was created through random sampling and combining of proteins of the positive set. The final, complete dataset was used to train an SVM, as well as a Random Forest classifier. Both models appeared to yield very good prediction accuracy, as well as specificity and recall. The promising results of this method highlight the importance of protein domains in PPI prediction, showcasing domains as key players in proteinprotein interactions.

Language

English

Subject

Interactome

Machine learning

Random forest

SVM

Αλληλεπίδρωμα

Μηχανική μάθηση

Issue date

2022-11-25

Collection