{"updated":"2025-01-19T15:41:04.271247+00:00","links":{},"created":"2025-01-19T01:17:26.309209+00:00","metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00216927","sets":["1164:5352:10882:10883"]},"path":["10883"],"owner":"44499","recid":"216927","title":["任意の分子を出発点とするグラフベースの分子最適化手法の開発"],"pubdate":{"attribute_name":"公開日","attribute_value":"2022-03-03"},"_buckets":{"deposit":"f3bb0765-3d68-4671-94dd-13ce7d94631d"},"_deposit":{"id":"216927","pid":{"type":"depid","value":"216927","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"任意の分子を出発点とするグラフベースの分子最適化手法の開発","author_link":["560990","560991","560988","560989","560992","560987"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"任意の分子を出発点とするグラフベースの分子最適化手法の開発"},{"subitem_title":"Development of a graph-based molecular optimization method with arbitrary molecules as a starting point","subitem_title_language":"en"}]},"item_type_id":"4","publish_date":"2022-03-03","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"東京工業大学情報理工学院"},{"subitem_text_value":"東京工業大学物質・情報卓越教育院"},{"subitem_text_value":"東京工業大学情報理工学院／東京工業大学物質・情報卓越教育院"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"jpn"}]},"item_publisher":{"attribute_name":"出版者","attribute_value_mlt":[{"subitem_publisher":"情報処理学会","subitem_publisher_language":"ja"}]},"publish_status":"0","weko_shared_id":-1,"item_file_price":{"attribute_name":"Billing file","attribute_type":"file","attribute_value_mlt":[{"url":{"url":"https://ipsj.ixsq.nii.ac.jp/record/216927/files/IPSJ-BIO22069017.pdf","label":"IPSJ-BIO22069017.pdf"},"date":[{"dateType":"Available","dateValue":"2024-03-03"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-BIO22069017.pdf","filesize":[{"value":"1.2 MB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"660","billingrole":"5"},{"tax":["include_tax"],"price":"330","billingrole":"6"},{"tax":["include_tax"],"price":"0","billingrole":"41"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"57f91658-04ad-40fc-9bc2-4dc6b75f8bd4","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2022 by the Information Processing Society of Japan"}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"恵利川, 大樹"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"安尾, 信明"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"関嶋, 政和"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Daiki, Erikawa","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Nobuaki, Yasuo","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Masakazu, Sekijima","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_4_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AA12055912","subitem_source_identifier_type":"NCID"}]},"item_4_textarea_12":{"attribute_name":"Notice","attribute_value_mlt":[{"subitem_textarea_value":"SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc."}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourceuri":"http://purl.org/coar/resource_type/c_18gh","resourcetype":"technical report"}]},"item_4_source_id_11":{"attribute_name":"ISSN","attribute_value_mlt":[{"subitem_source_identifier":"2188-8590","subitem_source_identifier_type":"ISSN"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"新薬を1つ開発するためには平均15年の期間と26憶ドルの費用を要する。そこで計算機科学の手法を用いてこの問題を解決する研究が取り組まれてきた。分子生成モデルはその1例であり、化合物データセットを学習することにより新しい化合物を生成する。本研究ではリード最適化のような出発点となる分子が与えられ評価関数に従い最適化する分子生成モデルの開発を行なった。この手法は分子グラフをベースとして、2つのモンテカルロ木探索でフラグメント単位の探索と原子単位の探索を行う。この時、グラフニューラルネットワークを同時に用いることにより効率的な探索を実現する。既存のフラグメント単位の生成手法の多くが固定されたフラグメントボキャブラリを用いているのに対して本手法は原子単位で随時フラグメントを生成するため多様性に優れている。実験では評価関数の1例としてよく用いられるQEDの最適化を行なった。結果として得られた化合物は比較手法よりも出発点の化合物と高い類似度を保ちながらQEDが上昇していることが確認できた。本手法は出発点の化合物の維持したい部分構造を指定できるなど柔軟な使い方が可能であり、創薬を始めとして様々な場面での利用が望まれる。","subitem_description_type":"Other"}]},"item_4_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"Development of a new drug takes a huge amount of time and money, on average 15 years and 2.6 billion dollars. Therefore, computational methods have been used to tackle this problem. Molecular generative models are one example, which generate new compounds by training compound data sets. In this study, we have developed a molecular generative model that optimizes a starting molecule according to an evaluation function. The method is based on a molecular graph and uses two Monte Carlo tree searches, one for fragments and the other for atoms. By using a graph neural network at the same time, efficient search is enabled. While most of the existing fragment generation methods use fixed fragment vocabularies, our method has high diversity because fragments are generated atom by atom. In the experiment, the QED was optimized, and the generated compound showed an increase in QED while maintaining a higher degree of similarity to the starting compound than the comparative method. This method allows the user to specify the substructure of the starting compound to be maintained. It is useful in a variety of situations including drug discovery.","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"6","bibliographic_titles":[{"bibliographic_title":"研究報告バイオ情報学（BIO）"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2022-03-03","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"17","bibliographicVolumeNumber":"2022-BIO-69"}]},"relation_version_is_last":true,"weko_creator_id":"44499"},"id":216927}