Single-Cell Techniques and Deep Learning in Predicting Drug Response - PubMed (original) (raw)

Review

Single-Cell Techniques and Deep Learning in Predicting Drug Response

Zhenyu Wu et al. Trends Pharmacol Sci. 2020 Dec.

Abstract

Rapidly developing single-cell sequencing analyses produce more comprehensive profiles of the genomic, transcriptomic, and epigenomic heterogeneity of tumor subpopulations than do traditional bulk sequencing analyses. Moreover, single-cell techniques allow the response of a tumor to drug exposure to be more thoroughlyinvestigated. Deep learning (DL) models have successfully extracted features from complex bulk sequence data to predict drug responses. We review recent innovations in single-cell technologies and DL-based approaches related to drug sensitivity predictions. We believe that, by using insights from bulk sequencedata, deep transfer learning (DTL) can facilitate the use of single-cell data for training superior DL-based drug prediction models.

Keywords: deep learning models; deep transfer learning framework; drug response; single-cell technologies.

Copyright © 2020 Elsevier Ltd. All rights reserved.

PubMed Disclaimer

Conflict of interest statement

Disclaimer Statement

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Figures

Figure 1.

Figure 1.. Key Figure. Combination of Single-cell and DL models in drug sensitivity prediction.

DL models are typically trained on tumor profiles, chemical and structural information of drugs and drug-target data. Extracting high-dimensional features through multi-layer perceptrons, DL models can infer drug-target interactions, purpose new drugs, and predict drug resistance.

Figure 2.

Figure 2.. Prediction of drug sensitivity at the single-cell level.

(A) Tumor subpopulations maintain diverse sensitivity to different drugs. Single usage of a drug may obtain less treatment efficiency. Knowing drug sensitivity at the single-cell level can guide the development of combination treatment that maximizes the efficiency of killing tumor cells while minimizes damage to healthy cells. (B) The MRD will proliferate and differentiate into a new tumor population which induces cancer relapse. The understanding of specific signatures characterized in MRD cells can help to discover novel drugs that specifically target MRD. MRD-targeted drug(s) administered in combination with conventional treatments can cure cancer and prevent relapse.

Figure 3.

Figure 3.. Potential applications of DTL framework on single-cell data for drug sensitivity prediction.

(A) the combination usage of generative adversarial network and DTL framework transfers the drug sensitivity known at the bulk level to the single-cell level. (B) A more advanced application of DTL would transfer drug sensitivity between two single-cell data and use bulk level information as a regularizer to constrain the DL parameters.

Box 1, Figure I.

Box 1, Figure I.. The scheme of deep learning models and frameworks.

(A) Four deep learning models that have been applied in drug prediction, including DNN, CNN, RNN, and GCN. (B) Two frameworks that can be combined with DL models.

Similar articles

Cited by

References

    1. Siegfried Z and Karni R (2018) The role of alternative splicing in cancer drug resistance. Current Opinion in Genetics & Development 48, 16–21. - PubMed
    1. Lee YT et al. (2018) Molecular targeted therapy: Treating cancer with specificity. European Journal of Pharmacology 834, 188–196. - PubMed
    1. Dagogo-Jack I and Shaw AT (2018) Tumour heterogeneity and resistance to cancer therapies. Nature Reviews Clinical Oncology 15 (2), 81–94. - PubMed
    1. Luskin MR et al. (2018) Targeting minimal residual disease: a path to cure? Nature reviews. Cancer 18 (4), 255–263. - PMC - PubMed
    1. Konieczkowski DJ et al. (2018) A Convergence-Based Framework for Cancer Drug Resistance. Cancer cell 33 (5), 801–815. - PMC - PubMed

Publication types

MeSH terms

Substances

LinkOut - more resources