Identification of Novel Antifibrotic Drug by Data-Driven Research Using AI

2022 Fiscal Year Final Research Report

• Project/Area Number: 20K15422
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 37030:Chemical biology-related
• Research Institution: Osaka University
• Principal Investigator: Nojima Yosui 大阪大学, 数理・データ科学教育研究センター, 特任講師(常勤) (30815717)
• Project Period (FY): 2020-04-01 – 2023-03-31
• Keywords: 人工知能 / AI / 機械学習 / IPF / ドラッグリポジショニング / 計算生物学 / 創薬探索 / マルチオミクス

Outline of Final Research Achievements

In the first year, I aimed to develop an AI for predicting cell survival rates through compound addition. The evaluation was conducted using two types of cross-validation methods. In the following year, I inputted multiomics data and compound information from IPF (Idiopathic Pulmonary Fibrosis) patients into the model constructed in the previous year to predict cell survival rates for each compound. Originally, our plan was to purchase IPF patient samples and collect multiomics data, but due to the impact of the COVID-19 pandemic, it became difficult to obtain the samples. Therefore, I partially collected multiomics data from IPF patient lungs through public databases. As a result, it was found that integrating multiple omics data, such as genes, proteins, and metabolites, is essential for improving accuracy, as it fell below the accuracy obtained in the first year.

Free Research Field

生命科学、バイオインフォマティクス、計算生物学

Academic Significance and Societal Importance of the Research Achievements

IPFなどの難病はデータ量が少なく、IPF単独のデータでは人工知能による解析は不向きである。そこで本研究では、IPFと同じく細胞増殖によって発症し、情報が豊富にある癌のデータに着目した。単一の疾患では人工知能の入力データ量として不十分であっても、類似性のある疾患のデータを用いてカバーすることができれば、希少難病などの領域にも人工知能による解析アプローチが適用できると考えられる。