Prediction of protein druggability by supervised learning approaches
سال انتشار: 1401
نوع سند: مقاله کنفرانسی
زبان: انگلیسی
مشاهده: 117
نسخه کامل این مقاله ارائه نشده است و در دسترس نمی باشد
- صدور گواهی نمایه سازی
- من نویسنده این مقاله هستم
استخراج به نرم افزارهای پژوهشی:
شناسه ملی سند علمی:
IBIS11_009
تاریخ نمایه سازی: 19 آذر 1402
چکیده مقاله:
Drug discovery includes several steps that lead from a biological hypothesis to an approved drug. Target identifica tion is usually the starting point of the modern drug discovery process. Candidate targets may be selected basedon different experimental criteria . Nowadays, most druggable targets or drug receptor molecules are proteins. Asa different approach from traditional drug development, machine learning algorithms are used to predict drug tar gets using data mining. This method has received more attention in recent years due to its various advantages suchas reducing cost and time. The primary approach of this research is to provide appropriate models for extractingand selecting the features of proteins and predicting their druggability. The first step included data extraction insuch a way that druggable and non-druggable protein sequences were taken from valid databases such as Drug Bank and UniProt, and a balance was established between positive and negative data. Physiochemical featuressuch as amino acid composition (AAC), composition of k-spaced amino acid group pairs (CKSAAGP), Moran cor relation, etc. are extracted using appropriate methods and functions . In the next step, by the use of supportvector machine (SVM), k-nearest neighbors (KNN), multilayer perceptron (MLP), and other classification algo rithms, the belonging of each protein sequence to either the druggable or non-druggable group was determined.Results were measured based on various performance metrics. The best algorithm among the methods used wasSVM with a polynomial kernel, which obtained an accuracy value of ۹۰ percent, which indicates that the selectedfeatures could be a good substitute for the features obtained from previous studies. In conclusion, results obtainedfrom this method are promising and this approach could be expanded for more extensive data in future studies
کلیدواژه ها:
نویسندگان
Parvin mansouri
Alzahra university
mahboobeh zarrabi
Alzahra university.
fatemeh ebrahimi tarki
Alzahra university.