Methodology, performance and retrainability survey of intrinsic disorder predictors
Year: 2023
Type: conference-paper
Abstract: Intrinsically disordered proteins and regions are widely distributed within most proteomes. \nRecent studies show that they are associated with many essential biological processes \nand a broad range of human diseases. Given the prevalence of disordered proteins and \nthe growing acknowledgement of their functional relevance, considerable effort has been \nmade by the bioinformatics community to provide computational tools to predict protein \ndisorder. To date, based on various characteristics of protein disorder, along with variety of \ndiverse computational approaches, numerous disorder predictors have been developed. \nOver the past decade several review papers examining intrinsic disorder predictors have \nbeen published. All these papers have played a significant role in stimulating and greatly \nfacilitating the development of this actively growing field by pinpointing the potential \nroom for improvement. Inspired by these, in this work we aim to integrate the relevant \ninformation regarding the existing intrinsic disorder predictors from the corresponding \nresearch papers in a novel review, including latest prediction tools. In addition, for each \ndisorder predictor, we examined the possibility of their retraining using different datasets. \nHere, we present an overview of 23 protein disorder prediction methods, including the \nthorough analysis of their advantages and weaknesses which derive from their different \ncomputational approaches. Regarding this, we precisely describe the methodology used \nfor building the models and categorize them by different classification schemes. The \nperformance of these models is presented by their scores from the most recent CAID \ncompetition. Additional contribution of this work is the models’ retraining availability \nanalysis. We describe in detail the predictors’ implementation source code (if available) and \npropose a way around to overcome the obstacles with retraining procedure (if possible). \nThis insight might be very useful, since older models were trained on significantly \nsmaller datasets compared to the newer ones, due to the increase in the number of \nexperimentally annotated disorder proteins with time. With respect to this, we discuss \nin detail the possibility of retraining the models on a different (bigger, novel) dataset in \norder to perform full-scale comparison of their expression power in delineating disorder \nin proteins.
Authors: Ćirić, Nevena, Kovačević, Jovana
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W6983489109