AttDRP: 주의집중 메커니즘 기반의 항암제 약물 반응성 예측 모델 = AttDRP: Attention Mechanism-based Model for Anti-Cancer Drug Response Prediction

다국어 초록 (Multilingual Abstract)

Resistance to anti-cancer drugs makes chemotherapy ineffective for cancer patients.
Drug resistance is caused by genetic alterations in cancer cells. Many studies have investigated drug responses of diverse cancer cell lines to various anti-cancer drugs to understand drug response mechanisms. Existing studies have proposed machine learning models for drug response prediction to find effective anti-cancer drugs. However, currently there are no models to learn the relationship between anticancer drugs and genes to improve the prediction accuracy. In this paper, we proposed a predictive model AttDRP that could identify important genes associated with anti-cancer drugs and predict drug responses based on identified genes. AttDRP exhibited better predictive accuracy than existing models and we found that the attention scores of AttDRP could be effective tools to analyze molecular structures of anticancer drugs. We hope that our proposed method would contribute to the development of precision medicine for effective chemotherapy. Resistance to anti-cancer drugs makes chemotherapy ineffective for cancer patients. Drug resistance is caused by genetic alterations in cancer cells. Many studies have investigated drug responses of diverse cancer cell lines to various anti-cancer drugs to understand drug response mechanisms. Existing studies have proposed machine learning models for drug response prediction to find effective anti-cancer drugs. However, currently there are no models to learn the relationship between anticancer drugs and genes to improve the prediction accuracy. In this paper, we proposed a predictive model AttDRP that could identify important genes associated with anti-cancer drugs and predict drug responses based on identified genes. AttDRP exhibited better predictive accuracy than existing models and we found that the attention scores of AttDRP could be effective tools to analyze molecular structures of anticancer drugs.

국문 초록 (Abstract)

암환자 중 일부는 항암제에 대한 약물 저항성을 보여 약물을 이용한 항암치료를 어렵게 만든다. 약물 저항성은 암세포의 유전체 이상에 기인하는 것으로 밝혀져, 암세포주 및 항암제에 대한 약물 반응성 데이터를 분석하는 연구가 활발히 이루어지고 있다. 기존 연구들은 기계학습을 이용하여 약물 민감성 또는 저항성을 예측하는 모델을 여럿 제안하였으나, 항암제와 유전자의 관계를 학습하는 모델의 부재로 인하여 예측 정확도 향상을 위한 여지가 남아있었다. 본 논문에서는 주의집중 메커니즘을 활용하여 항암제 관련 유전자들을 식별하고, 그러한 유전자들 정보에 기반하여 항암제 반응성을 예측하는 AttDRP를 제안한다. 제안하는 모델은 CCLE 데이터에서 기존 모델들보다 높은 예측 정확도를 보여주었고, AttDRP이 학습한 주의집중 스코어가 항암제의 분자구조 분석에 효과적으로 활용될 수 있음을 확인하였다.

참고문헌 (Reference)

1 McInnes, Leland, "Umap: Uniform manifold approximation and projection for dimension reduction"

2 Barretina, Jordi, "The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity" 483 (483): 603-607, 2012

3 Park, Sang Min, "Systems analysis identifies potential target genes to overcome cetuximab resistance in colorectal cancer cells" 286 (286): 1305-1318, 2019

4 Choi, Jonghwan, "RefDNN: a reference drug based neural network for more accurate prediction of anticancer drug resistance" 10 (10): 1-11, 2020

5 "PubChem substructure fingerprint V1.3"

6 Suphavilai, Chayaporn, "Predicting cancer drug response using a recommender system" 34 (34): 3907-3914, 2018

7 Munkácsy, Gyöngyi, "PSMB7 is associated with anthracycline resistance and is a prognostic biomarker in breast cancer" 102 (102): 361-368, 2009

8 Rácz, Anita, "Life beyond the Tanimoto coefficient: similarity measures for interaction fingerprints" 10 (10): 1-12, 2018

9 Ammad-Ud-Din, Muhammad, "Integrative and personalized QSAR analysis in cancer by kernelized Bayesian matrix factorization" 54 (54): 2347-2359, 2014

10 Wang, Lin, "Improved anticancer drug response prediction in cell lines using matrix factorization with similarity regularization" 17 (17): 1-12, 2017

11 Yang, Wanjuan, "Genomics of Drug Sensitivity in Cancer (GDSC): a resource for therapeutic biomarker discovery in cancer cells" 41 (41): D955-D961, 2013

12 Stanfield, Zachary, "Drug response prediction as a link prediction problem" 7 (7): 1-12, 2017

13 Housman, Genevieve, "Drug resistance in cancer: an overview" 6 (6): 1769-1792, 2014

14 Jiao, Xiaoli, "DAVID-WS: a stateful web service to facilitate gene/protein list analysis" 28 (28): 1805-1806, 2012

15 Zhang, Fei, "A novel heterogeneous network-based method for drug response prediction in cancer cell lines" 8 (8): 1-9, 2018

16 Costello, James C., "A community effort to assess and improve drug sensitivity prediction algorithms" 32 (32): 1202-1212, 2014