1. Protein folding using deep reinforcement learning
- Subjects
ma- chine learning ,глÑбокое обÑÑение ,reinforcement learning ,обÑÑение Ñ Ð¿Ð¾Ð´ÐºÑеплением ,protein folding ,маÑинное обÑÑение ,deep learning ,optimization task ,Ñолдинг белка ,задаÑа опÑимизаÑии - Abstract
Ð ÑабоÑе пÑедложен меÑод ÑеÑÐµÐ½Ð¸Ñ Ð·Ð°Ð´Ð°Ñи Ñолдинга белка пÑи помоÑи глÑбокого обÑÑÐµÐ½Ð¸Ñ Ñ Ð¿Ð¾Ð´ÐºÑеплением. Создана маÑемаÑиÑеÑÐºÐ°Ñ Ð¼Ð¾Ð´ÐµÐ»Ñ ÑизиÑеÑÐºÐ¸Ñ Ð¿ÑоÑеÑÑов пÑи ÑвоÑаÑивании гидÑоÑобно-гидÑоÑилÑного (HP-моделÑ) белка. ÐолÑÑÐµÐ½Ñ ÑезÑлÑÑаÑÑ ÑеÑÐµÐ½Ð¸Ñ Ð¾Ð¿ÑимизаÑÐ¸Ð¾Ð½Ð½Ð¾Ð¸Ì Ð·Ð°Ð´Ð°Ñи, коÑоÑÑе показÑваÑÑ ÐºÐ°Ðº колиÑеÑÑвенно близкие к опÑималÑÐ½Ð¾Ð¼Ñ ÑеÑÐµÐ½Ð¸Ñ ÑпоÑÐ¾Ð±Ñ ÑвоÑаÑÐ¸Ð²Ð°Ð½Ð¸Ñ Ð±ÐµÐ»ÐºÐ°, Ñак и каÑеÑÑвенно обоÑнованнÑе Ð±Ð¸Ð¾Ñ Ð¸Ð¼Ð¸ÑеÑÐºÐ¾Ð¸Ì Ð¸ биоÑизиÑеÑÐºÐ¾Ð¸Ì ÑеоÑиеиÌ., The paper proposes a method for solving the protein folding problem using deep reinforcement learning. A mathematical model of physical processes during the folding of hydrophobic-hydrophilic (HP-model) protein has been created. The results of solving the optimization problem are obtained, which show both quantitatively close to the optimal solution methods of protein folding, and qualitatively justified by biochemical and biophysical theory.
- Published
- 2022
- Full Text
- View/download PDF