Loading…

PaccMann RL : De novo generation of hit-like anticancer molecules from transcriptomic data via reinforcement learning

With the advent of deep generative models in computational chemistry, drug design is undergoing an unprecedented transformation. Although deep learning approaches have shown potential in generating compounds with desired chemical properties, they disregard the cellular environment of target diseases...

Full description

Saved in:
Bibliographic Details
Published in:iScience 2021-04, Vol.24 (4), p.102269
Main Authors: Born, Jannis, Manica, Matteo, Oskooei, Ali, Cadow, Joris, Markert, Greta, Rodríguez Martínez, María
Format: Article
Language:English
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
cited_by
cites
container_end_page
container_issue 4
container_start_page 102269
container_title iScience
container_volume 24
creator Born, Jannis
Manica, Matteo
Oskooei, Ali
Cadow, Joris
Markert, Greta
Rodríguez Martínez, María
description With the advent of deep generative models in computational chemistry, drug design is undergoing an unprecedented transformation. Although deep learning approaches have shown potential in generating compounds with desired chemical properties, they disregard the cellular environment of target diseases. Bridging systems biology and drug design, we present a reinforcement learning method for de novo molecular design from gene expression profiles. We construct a hybrid Variational Autoencoder that tailors molecules to target-specific transcriptomic profiles, using an anticancer drug sensitivity prediction model (PaccMann) as reward function. Without incorporating information about anticancer drugs, the molecule generation is biased toward compounds with high predicted efficacy against cell lines or cancer types. The generation can be further refined by subsidiary constraints such as toxicity. Our cancer-type-specific candidate drugs are similar to cancer drugs in drug-likeness, synthesizability, and solubility and frequently exhibit the highest structural similarity to compounds with known efficacy against these cancer types.
format article
fullrecord <record><control><sourceid>pubmed</sourceid><recordid>TN_cdi_pubmed_primary_33851095</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>33851095</sourcerecordid><originalsourceid>FETCH-pubmed_primary_338510953</originalsourceid><addsrcrecordid>eNqFjsFqAjEQQEOhqFR_ocwPLMRdV1yvVvFQoRTvMsZZHZtMlklW8O_10J4LD97lHd6LGZX1oimsnZVDM0npaq0tn8ya-cAMq2pRT21Tj0z_hc7tUAS-P2EJHwQSbxHOJKSYOQrEFi6cC88_BCiZHYojhRA9ud5TglZjgKwoySl3OQZ2cMKMcGMEJZY2qqNAksETqrCcx-a1RZ9o8us3875Z71fbouuPgU6HTjmg3g9_n9W_wQOpTkqh</addsrcrecordid><sourcetype>Index Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>PaccMann RL : De novo generation of hit-like anticancer molecules from transcriptomic data via reinforcement learning</title><source>Open Access: PubMed Central</source><source>ScienceDirect (Online service)</source><creator>Born, Jannis ; Manica, Matteo ; Oskooei, Ali ; Cadow, Joris ; Markert, Greta ; Rodríguez Martínez, María</creator><creatorcontrib>Born, Jannis ; Manica, Matteo ; Oskooei, Ali ; Cadow, Joris ; Markert, Greta ; Rodríguez Martínez, María</creatorcontrib><description>With the advent of deep generative models in computational chemistry, drug design is undergoing an unprecedented transformation. Although deep learning approaches have shown potential in generating compounds with desired chemical properties, they disregard the cellular environment of target diseases. Bridging systems biology and drug design, we present a reinforcement learning method for de novo molecular design from gene expression profiles. We construct a hybrid Variational Autoencoder that tailors molecules to target-specific transcriptomic profiles, using an anticancer drug sensitivity prediction model (PaccMann) as reward function. Without incorporating information about anticancer drugs, the molecule generation is biased toward compounds with high predicted efficacy against cell lines or cancer types. The generation can be further refined by subsidiary constraints such as toxicity. Our cancer-type-specific candidate drugs are similar to cancer drugs in drug-likeness, synthesizability, and solubility and frequently exhibit the highest structural similarity to compounds with known efficacy against these cancer types.</description><identifier>EISSN: 2589-0042</identifier><identifier>PMID: 33851095</identifier><language>eng</language><publisher>United States</publisher><ispartof>iScience, 2021-04, Vol.24 (4), p.102269</ispartof><rights>2021 The Author(s).</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>314,780,784</link.rule.ids><backlink>$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/33851095$$D View this record in MEDLINE/PubMed$$Hfree_for_read</backlink></links><search><creatorcontrib>Born, Jannis</creatorcontrib><creatorcontrib>Manica, Matteo</creatorcontrib><creatorcontrib>Oskooei, Ali</creatorcontrib><creatorcontrib>Cadow, Joris</creatorcontrib><creatorcontrib>Markert, Greta</creatorcontrib><creatorcontrib>Rodríguez Martínez, María</creatorcontrib><title>PaccMann RL : De novo generation of hit-like anticancer molecules from transcriptomic data via reinforcement learning</title><title>iScience</title><addtitle>iScience</addtitle><description>With the advent of deep generative models in computational chemistry, drug design is undergoing an unprecedented transformation. Although deep learning approaches have shown potential in generating compounds with desired chemical properties, they disregard the cellular environment of target diseases. Bridging systems biology and drug design, we present a reinforcement learning method for de novo molecular design from gene expression profiles. We construct a hybrid Variational Autoencoder that tailors molecules to target-specific transcriptomic profiles, using an anticancer drug sensitivity prediction model (PaccMann) as reward function. Without incorporating information about anticancer drugs, the molecule generation is biased toward compounds with high predicted efficacy against cell lines or cancer types. The generation can be further refined by subsidiary constraints such as toxicity. Our cancer-type-specific candidate drugs are similar to cancer drugs in drug-likeness, synthesizability, and solubility and frequently exhibit the highest structural similarity to compounds with known efficacy against these cancer types.</description><issn>2589-0042</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</creationdate><recordtype>article</recordtype><recordid>eNqFjsFqAjEQQEOhqFR_ocwPLMRdV1yvVvFQoRTvMsZZHZtMlklW8O_10J4LD97lHd6LGZX1oimsnZVDM0npaq0tn8ya-cAMq2pRT21Tj0z_hc7tUAS-P2EJHwQSbxHOJKSYOQrEFi6cC88_BCiZHYojhRA9ud5TglZjgKwoySl3OQZ2cMKMcGMEJZY2qqNAksETqrCcx-a1RZ9o8us3875Z71fbouuPgU6HTjmg3g9_n9W_wQOpTkqh</recordid><startdate>20210423</startdate><enddate>20210423</enddate><creator>Born, Jannis</creator><creator>Manica, Matteo</creator><creator>Oskooei, Ali</creator><creator>Cadow, Joris</creator><creator>Markert, Greta</creator><creator>Rodríguez Martínez, María</creator><scope>NPM</scope></search><sort><creationdate>20210423</creationdate><title>PaccMann RL : De novo generation of hit-like anticancer molecules from transcriptomic data via reinforcement learning</title><author>Born, Jannis ; Manica, Matteo ; Oskooei, Ali ; Cadow, Joris ; Markert, Greta ; Rodríguez Martínez, María</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-pubmed_primary_338510953</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Born, Jannis</creatorcontrib><creatorcontrib>Manica, Matteo</creatorcontrib><creatorcontrib>Oskooei, Ali</creatorcontrib><creatorcontrib>Cadow, Joris</creatorcontrib><creatorcontrib>Markert, Greta</creatorcontrib><creatorcontrib>Rodríguez Martínez, María</creatorcontrib><collection>PubMed</collection><jtitle>iScience</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Born, Jannis</au><au>Manica, Matteo</au><au>Oskooei, Ali</au><au>Cadow, Joris</au><au>Markert, Greta</au><au>Rodríguez Martínez, María</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>PaccMann RL : De novo generation of hit-like anticancer molecules from transcriptomic data via reinforcement learning</atitle><jtitle>iScience</jtitle><addtitle>iScience</addtitle><date>2021-04-23</date><risdate>2021</risdate><volume>24</volume><issue>4</issue><spage>102269</spage><pages>102269-</pages><eissn>2589-0042</eissn><abstract>With the advent of deep generative models in computational chemistry, drug design is undergoing an unprecedented transformation. Although deep learning approaches have shown potential in generating compounds with desired chemical properties, they disregard the cellular environment of target diseases. Bridging systems biology and drug design, we present a reinforcement learning method for de novo molecular design from gene expression profiles. We construct a hybrid Variational Autoencoder that tailors molecules to target-specific transcriptomic profiles, using an anticancer drug sensitivity prediction model (PaccMann) as reward function. Without incorporating information about anticancer drugs, the molecule generation is biased toward compounds with high predicted efficacy against cell lines or cancer types. The generation can be further refined by subsidiary constraints such as toxicity. Our cancer-type-specific candidate drugs are similar to cancer drugs in drug-likeness, synthesizability, and solubility and frequently exhibit the highest structural similarity to compounds with known efficacy against these cancer types.</abstract><cop>United States</cop><pmid>33851095</pmid></addata></record>
fulltext fulltext
identifier EISSN: 2589-0042
ispartof iScience, 2021-04, Vol.24 (4), p.102269
issn 2589-0042
language eng
recordid cdi_pubmed_primary_33851095
source Open Access: PubMed Central; ScienceDirect (Online service)
title PaccMann RL : De novo generation of hit-like anticancer molecules from transcriptomic data via reinforcement learning
url http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-25T19%3A57%3A08IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-pubmed&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=PaccMann%20RL%20:%20De%20novo%20generation%20of%20hit-like%20anticancer%20molecules%20from%20transcriptomic%20data%20via%20reinforcement%20learning&rft.jtitle=iScience&rft.au=Born,%20Jannis&rft.date=2021-04-23&rft.volume=24&rft.issue=4&rft.spage=102269&rft.pages=102269-&rft.eissn=2589-0042&rft_id=info:doi/&rft_dat=%3Cpubmed%3E33851095%3C/pubmed%3E%3Cgrp_id%3Ecdi_FETCH-pubmed_primary_338510953%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_id=info:pmid/33851095&rfr_iscdi=true