ReSimNet: drug response similarity prediction using Siamese neural networks

Minji Jeon, Donghyeon Park, Jinhyuk Lee, Hwisang Jeon, Miyoung Ko, Sunkyu Kim, Yonghwa Choi, Aik Choon Tan, Jaewoo Kang

Research output: Contribution to journalArticle

Abstract

MOTIVATION: Traditional drug discovery approaches identify a target for a disease and find a compound that binds to the target. In this approach, structures of compounds are considered as the most important features because it is assumed that similar structures will bind to the same target. Therefore, structural analogs of the drugs that bind to the target are selected as drug candidates. However, even though compounds are not structural analogs, they may achieve the desired response. A new drug discovery method based on drug response, which can complement the structure-based methods, is needed. RESULTS: We implemented Siamese neural networks called ReSimNet that take as input two chemical compounds and predicts the CMap score of the two compounds, which we use to measure the transcriptional response similarity of the two compounds. ReSimNet learns the embedding vector of a chemical compound in a transcriptional response space. ReSimNet is trained to minimize the difference between the cosine similarity of the embedding vectors of the two compounds and the CMap score of the two compounds. ReSimNet can find pairs of compounds that are similar in response even though they may have dissimilar structures. In our quantitative evaluation, ReSimNet outperformed the baseline machine learning models. The ReSimNet ensemble model achieves a Pearson correlation of 0.518 and a precision@1% of 0.989. In addition, in the qualitative analysis, we tested ReSimNet on the ZINC15 database and showed that ReSimNet successfully identifies chemical compounds that are relevant to a prototype drug whose mechanism of action is known. AVAILABILITY AND IMPLEMENTATION: The source code and the pre-trained weights of ReSimNet are available at https://github.com/dmis-lab/ReSimNet. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Original languageEnglish
Pages (from-to)5249-5256
Number of pages8
JournalBioinformatics (Oxford, England)
Volume35
Issue number24
DOIs
Publication statusPublished - 2019 Dec 15

Fingerprint

Chemical compounds
Drugs
Neural Networks
Neural networks
Prediction
Drug Discovery
Target
Pharmaceutical Preparations
Bioinformatics
Learning systems
Analogue
Computational Biology
Pearson Correlation
Availability
Quantitative Evaluation
Qualitative Analysis
Databases
Baseline
Machine Learning
Weights and Measures

ASJC Scopus subject areas

  • Statistics and Probability
  • Biochemistry
  • Molecular Biology
  • Computer Science Applications
  • Computational Theory and Mathematics
  • Computational Mathematics

Cite this

ReSimNet : drug response similarity prediction using Siamese neural networks. / Jeon, Minji; Park, Donghyeon; Lee, Jinhyuk; Jeon, Hwisang; Ko, Miyoung; Kim, Sunkyu; Choi, Yonghwa; Tan, Aik Choon; Kang, Jaewoo.

In: Bioinformatics (Oxford, England), Vol. 35, No. 24, 15.12.2019, p. 5249-5256.

Research output: Contribution to journalArticle

Jeon, M, Park, D, Lee, J, Jeon, H, Ko, M, Kim, S, Choi, Y, Tan, AC & Kang, J 2019, 'ReSimNet: drug response similarity prediction using Siamese neural networks', Bioinformatics (Oxford, England), vol. 35, no. 24, pp. 5249-5256. https://doi.org/10.1093/bioinformatics/btz411
Jeon, Minji ; Park, Donghyeon ; Lee, Jinhyuk ; Jeon, Hwisang ; Ko, Miyoung ; Kim, Sunkyu ; Choi, Yonghwa ; Tan, Aik Choon ; Kang, Jaewoo. / ReSimNet : drug response similarity prediction using Siamese neural networks. In: Bioinformatics (Oxford, England). 2019 ; Vol. 35, No. 24. pp. 5249-5256.
@article{fb1e4cd5b6694ff199ecff3ae572cfc9,
title = "ReSimNet: drug response similarity prediction using Siamese neural networks",
abstract = "MOTIVATION: Traditional drug discovery approaches identify a target for a disease and find a compound that binds to the target. In this approach, structures of compounds are considered as the most important features because it is assumed that similar structures will bind to the same target. Therefore, structural analogs of the drugs that bind to the target are selected as drug candidates. However, even though compounds are not structural analogs, they may achieve the desired response. A new drug discovery method based on drug response, which can complement the structure-based methods, is needed. RESULTS: We implemented Siamese neural networks called ReSimNet that take as input two chemical compounds and predicts the CMap score of the two compounds, which we use to measure the transcriptional response similarity of the two compounds. ReSimNet learns the embedding vector of a chemical compound in a transcriptional response space. ReSimNet is trained to minimize the difference between the cosine similarity of the embedding vectors of the two compounds and the CMap score of the two compounds. ReSimNet can find pairs of compounds that are similar in response even though they may have dissimilar structures. In our quantitative evaluation, ReSimNet outperformed the baseline machine learning models. The ReSimNet ensemble model achieves a Pearson correlation of 0.518 and a precision@1{\%} of 0.989. In addition, in the qualitative analysis, we tested ReSimNet on the ZINC15 database and showed that ReSimNet successfully identifies chemical compounds that are relevant to a prototype drug whose mechanism of action is known. AVAILABILITY AND IMPLEMENTATION: The source code and the pre-trained weights of ReSimNet are available at https://github.com/dmis-lab/ReSimNet. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.",
author = "Minji Jeon and Donghyeon Park and Jinhyuk Lee and Hwisang Jeon and Miyoung Ko and Sunkyu Kim and Yonghwa Choi and Tan, {Aik Choon} and Jaewoo Kang",
year = "2019",
month = "12",
day = "15",
doi = "10.1093/bioinformatics/btz411",
language = "English",
volume = "35",
pages = "5249--5256",
journal = "Bioinformatics",
issn = "1367-4803",
publisher = "Oxford University Press",
number = "24",

}

TY - JOUR

T1 - ReSimNet

T2 - drug response similarity prediction using Siamese neural networks

AU - Jeon, Minji

AU - Park, Donghyeon

AU - Lee, Jinhyuk

AU - Jeon, Hwisang

AU - Ko, Miyoung

AU - Kim, Sunkyu

AU - Choi, Yonghwa

AU - Tan, Aik Choon

AU - Kang, Jaewoo

PY - 2019/12/15

Y1 - 2019/12/15

N2 - MOTIVATION: Traditional drug discovery approaches identify a target for a disease and find a compound that binds to the target. In this approach, structures of compounds are considered as the most important features because it is assumed that similar structures will bind to the same target. Therefore, structural analogs of the drugs that bind to the target are selected as drug candidates. However, even though compounds are not structural analogs, they may achieve the desired response. A new drug discovery method based on drug response, which can complement the structure-based methods, is needed. RESULTS: We implemented Siamese neural networks called ReSimNet that take as input two chemical compounds and predicts the CMap score of the two compounds, which we use to measure the transcriptional response similarity of the two compounds. ReSimNet learns the embedding vector of a chemical compound in a transcriptional response space. ReSimNet is trained to minimize the difference between the cosine similarity of the embedding vectors of the two compounds and the CMap score of the two compounds. ReSimNet can find pairs of compounds that are similar in response even though they may have dissimilar structures. In our quantitative evaluation, ReSimNet outperformed the baseline machine learning models. The ReSimNet ensemble model achieves a Pearson correlation of 0.518 and a precision@1% of 0.989. In addition, in the qualitative analysis, we tested ReSimNet on the ZINC15 database and showed that ReSimNet successfully identifies chemical compounds that are relevant to a prototype drug whose mechanism of action is known. AVAILABILITY AND IMPLEMENTATION: The source code and the pre-trained weights of ReSimNet are available at https://github.com/dmis-lab/ReSimNet. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

AB - MOTIVATION: Traditional drug discovery approaches identify a target for a disease and find a compound that binds to the target. In this approach, structures of compounds are considered as the most important features because it is assumed that similar structures will bind to the same target. Therefore, structural analogs of the drugs that bind to the target are selected as drug candidates. However, even though compounds are not structural analogs, they may achieve the desired response. A new drug discovery method based on drug response, which can complement the structure-based methods, is needed. RESULTS: We implemented Siamese neural networks called ReSimNet that take as input two chemical compounds and predicts the CMap score of the two compounds, which we use to measure the transcriptional response similarity of the two compounds. ReSimNet learns the embedding vector of a chemical compound in a transcriptional response space. ReSimNet is trained to minimize the difference between the cosine similarity of the embedding vectors of the two compounds and the CMap score of the two compounds. ReSimNet can find pairs of compounds that are similar in response even though they may have dissimilar structures. In our quantitative evaluation, ReSimNet outperformed the baseline machine learning models. The ReSimNet ensemble model achieves a Pearson correlation of 0.518 and a precision@1% of 0.989. In addition, in the qualitative analysis, we tested ReSimNet on the ZINC15 database and showed that ReSimNet successfully identifies chemical compounds that are relevant to a prototype drug whose mechanism of action is known. AVAILABILITY AND IMPLEMENTATION: The source code and the pre-trained weights of ReSimNet are available at https://github.com/dmis-lab/ReSimNet. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

UR - http://www.scopus.com/inward/record.url?scp=85077766813&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85077766813&partnerID=8YFLogxK

U2 - 10.1093/bioinformatics/btz411

DO - 10.1093/bioinformatics/btz411

M3 - Article

C2 - 31116384

AN - SCOPUS:85077766813

VL - 35

SP - 5249

EP - 5256

JO - Bioinformatics

JF - Bioinformatics

SN - 1367-4803

IS - 24

ER -