Please use this identifier to cite or link to this item:
Title: PPRINT: Prediction of Protein-Protein Interactions
Authors: Cruz, Igor Nelson Garrido da 
Orientador: Arrais, Joel Perdiz
Keywords: Bioinformatics; Protein Interaction Prediction; Protein Features; Machine Learning
Issue Date: 15-Sep-2014
Serial title, monograph or event: PPRINT: Prediction of Protein-Protein Interactions
Place of publication or event: Coimbra
Abstract: Understanding life at a molecular level, while complex, encloses a myriad of opportunities for humanity's future. As important as being able to identify the molecular components of the cell it is of major relevance to understand their relationships and interactions. This way, the study of Protein-Protein Interactions (PPIs) has been used as a cornerstone to de- termine how most of the biological processes take place. Due to the large scale of the problem it is critical to use the appropriate computational tools and methods. Despite the existence of previous works in the eld, the available methods are divided in two groups of approaches: experi- mental and computational. Experimental techniques have good prediction accuracy but are slow and expensive, therefore urges the need of develop- ing computational approaches. These have low prediction accuracy but only require computational power and consequently are inexpensive since no laboratory machinery is required. A great amount of these algorithms are based on protein annotations, such as protein homology or protein do- mains. That makes such algorithms inapplicable to sparse multi-organism datasets usually composed only by the proteins sequences. In this work we start by analysing the existent state of the art methods for compu- tational prediction of PPIs. It is our goal to explore their limitations and make improvements that can lead to more accurate results. After that we propose a new approach using the discrete cosine transform as a method of construction of features from the protein chain and a new method that calculates the three dimensional structure of the protein from its sequence. These new improved approaches will bequeath more accurate protein interactomes that can be used by Genomic Engineers in order to understand the intracellular structures relationship and biological processes. From these biological processes it is possible to extract seman- tic knowledge that can lead to new drug discoveries. Finally as a method of validation, our work is currently being experimentally validated by the Faculty of Dental Medicine from the Catholic University of Portugal from the biological perspective using real sets of proteins extracted from hu- mans saliva and from microorganisms presents in the oral cavity. It is also publicly available online for everyone to complement or use in other researches.
Description: Dissertação de Mestrado em Engenharia Informática apresentada à Faculdade de Ciências e Tecnologia da Universidade de Coimbra
Rights: openAccess
Appears in Collections:UC - Dissertações de Mestrado
FCTUC Eng.Informática - Teses de Mestrado

Files in This Item:
File Description SizeFormat
PPRINT Prediction of Protein Protein Interactions.pdf3.49 MBAdobe PDFView/Open
Show full item record

Page view(s) 20

checked on Oct 22, 2020


checked on Oct 22, 2020

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.