CoChat
Sign Up
Cited by 234Computational Drug Discovery MethodsMicrobial Natural Products and BiosynthesisAntimicrobial Peptides and Activities

The DOI matches its Crossref record; 208 of the 208 references it cites were found in a registry or an index.

Accelerating antibiotic discovery through artificial intelligence

Open access
Melo, Marcelo C. R., Maasch, Jacqueline R. M. A., de la Fuente-Nunez, Cesar·2021·Communications Biology
DOI 10.1038/s42003-021-02586-0PMC PMC8429579
Open this paper in CoChat — free accountOpen PDF

Abstract

By targeting invasive organisms, antibiotics insert themselves into the ancient struggle of the host-pathogen evolutionary arms race. As pathogens evolve tactics for evading antibiotics, therapies decline in efficacy and must be replaced, distinguishing antibiotics from most other forms of drug development. Together with a slow and expensive antibiotic development pipeline, the proliferation of drug-resistant pathogens drives urgent interest in computational methods that promise to expedite candidate discovery. Strides in artificial intelligence (AI) have encouraged its application to multiple dimensions of computer-aided drug design, with increasing application to antibiotic discovery. This review describes AI-facilitated advances in the discovery of both small molecule antibiotics and antimicrobial peptides. Beyond the essential prediction of antimicrobial activity, emphasis is also given to antimicrobial compound representation, determination of drug-likeness traits, antimicrobial resistance, and de novo molecular design. Given the urgency of the antimicrobial resistance crisis, we analyze uptake of open science best practices in AI-driven antibiotic discovery and argue for openness and reproducibility as a means of accelerating preclinical research. Finally, trends in the literature and areas for future inquiry are discussed, as artificially intelligent enhancements to drug discovery at large offer many opportunities for future applications in antibiotic development.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s42003-021-02586-0

Cites

208 of 208 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 Computational antibiotic discovery pipeline. The figure provides an overview of data and methods used in antibiotic discovery and development using AI. From left to right, key elements in the drug development process are exemplified. The first part of any AI-driven project is gathering the experimental information that will enable model creation. The data are then transformed into AI-ready representations. Subsequently, models are trained using algorithms that can range from traditional decision trees to novel neural networks. Finally, trained models can be used to predict diverse qualities, e.g., the effectiveness of an antibiotic, potential for toxic activity, development of resistance, or the structure of novel compounds that exhibit desirable traits.
Fig. 1 Computational antibiotic discovery pipeline. The figure provides an overview of data and methods used in antibiotic discovery and development using AI. From left to right, key elements in the drug development process are exemplified. The first part of any AI-driven project is gathering the experimental information that will enable model creation. The data are then transformed into AI-ready representations. Subsequently, models are trained using algorithms that can range from traditional decision trees to novel neural networks. Finally, trained models can be used to predict diverse qualities, e.g., the effectiveness of an antibiotic, potential for toxic activity, development of resistance, or the structure of novel compounds that exhibit desirable traits.CC BY · PubMed Central
Fig. 2 Open science practices in machine learning for antibiotic discovery. This Euler diagram visualizes public release rates for source code, training or testing data, software, and combinations thereof among publications cited in this review (Table 2). Note that data release criteria for this analysis include both partial and full public availability. This analysis was performed post hoc on studies previously cited in this review.
Fig. 2 Open science practices in machine learning for antibiotic discovery. This Euler diagram visualizes public release rates for source code, training or testing data, software, and combinations thereof among publications cited in this review (Table 2). Note that data release criteria for this analysis include both partial and full public availability. This analysis was performed post hoc on studies previously cited in this review.CC BY · PubMed Central
Fig. 3 Machine learning in antibiotic discovery over time. From top to bottom: total PubMed results when querying for AI/ML keywords only, total results when querying for AI/ML and general or disease group-specific drug keywords, and the proportion of general AI/ML publications pertaining to each category of drugs (i.e., total publication counts per drug category scaled by total AI/ML publications per year). Queries sought keywords in titles and abstracts only, with the general drug query excluding keywords contained in the disease group queries to prevent double-counting. Key events in the broader ML community are noted to contextualize trend lines. The relevant literature used to set key dates are as follows: development of SVM146 and random forest algorithms147 in 1995; publication of the R language and software environment in 1996148; development of LSTM in 1997149; development of the Biopython package in 2000150; release of the Java interface for Weka in 2002151; publication of the Torch library in 2002152; release of Bioconductor in 2004153; the publication of ImageNet in 2009154; the initial release of Scikit-learn in 2010155; the initial release of XGBoost156 and development of GANs102 in 2014; development of Keras157 and TensorFlow158 in 2015; and the initial release of PyTorch in 2016159. Exact Boolean searches in PubMed can be found in Supplementary Table 1.
Fig. 3 Machine learning in antibiotic discovery over time. From top to bottom: total PubMed results when querying for AI/ML keywords only, total results when querying for AI/ML and general or disease group-specific drug keywords, and the proportion of general AI/ML publications pertaining to each category of drugs (i.e., total publication counts per drug category scaled by total AI/ML publications per year). Queries sought keywords in titles and abstracts only, with the general drug query excluding keywords contained in the disease group queries to prevent double-counting. Key events in the broader ML community are noted to contextualize trend lines. The relevant literature used to set key dates are as follows: development of SVM146 and random forest algorithms147 in 1995; publication of the R language and software environment in 1996148; development of LSTM in 1997149; development of the Biopython package in 2000150; release of the Java interface for Weka in 2002151; publication of the Torch library in 2002152; release of Bioconductor in 2004153; the publication of ImageNet in 2009154; the initial release of Scikit-learn in 2010155; the initial release of XGBoost156 and development of GANs102 in 2014; development of Keras157 and TensorFlow158 in 2015; and the initial release of PyTorch in 2016159. Exact Boolean searches in PubMed can be found in Supplementary Table 1.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 208 references

Research works printed but not linked: 2

  1. Goodfellow, I. et al. Generative adversarial nets. In Proc. 2014 Advances in Neural Information Processing Systems 2672–2680 (2014).
  2. Corsello, S. M. et al. Inf. Resour. 23, 405–408 (2017).

Cited by

Sorted by citations

See all 234

Related

Metadata

Authors
Melo, M. C. R., Maasch, J. R. M. A., & de la Fuente-Nunez, C.
Affiliations
University of Pennsylvania (US)
ORCID
Venue
Communications Biology
Publisher
Springer Science and Business Media LLC
Published
2021-09-09
Volume
4
Issue
1
Pages
1050-1050
ISSN
2399-3642 (ISSN-L)
DOI
10.1038/s42003-021-02586-0
PMC
PMC8429579
OpenAlex
W3197063311
Cited by
234
References
208
Citations by year
2026: 44 · 2025: 60 · 2024: 65 · 2023: 42 · 2022: 20 · 2021: 2
FWCI
2.52
Type
journal article
Language
en
Open access
Licence
Locations
  • Communications Biology (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, other-oa)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
  • Alfred P. Sloan Foundation
  • University of Pennsylvania
  • Penn Mental Health AIDS Research Center
  • National Institutes of Health
  • Defense Threat Reduction Agency (HDTRA11810041, HDTRA12110014, R35GM138201, HDTRA1)
  • Perelman School of Medicine, University of Pennsylvania
  • National Institute of General Medical Sciences (HDTRA11810041, HDTRA1-21-1-0014, R35GM138201)
Topics
  • Life Sciences › Pharmacology, Toxicology and Pharmaceutics › Drug Discovery › Computational Drug Discovery Methods
  • Life Sciences › Immunology and Microbiology › Applied Microbiology and Biotechnology › Microbial Natural Products and Biosynthesis
  • Life Sciences › Immunology and Microbiology › Microbiology › Antimicrobial Peptides and Activities
Keywords
antibiotic discovery, artificial intelligence, antimicrobial resistance, antimicrobial peptides, small-molecule antibiotics, computer-aided drug design, antimicrobial activity prediction, drug-likeness, de novo molecular design, open science, reproducibility, host-pathogen arms race
MeSH terms
Machine Learning, Antimicrobial Peptides, Anti-Bacterial Agents, Artificial Intelligence, Drug Design, Reproducibility of Results, Drug Discovery

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review