Beam Search for Automated Design and Scoring of Novel ROR Ligands with Machine Intelligence**

Article abstract of DOI:10.1002/anie.202104405

Chemical language models enable de novo drug design without the requirement for explicit molecular construction rules. While such models have been applied to generate novel compounds with desired bioactivity, the actual prioritization and selection of the most promising computational designs remains challenging. Herein, we leveraged the probabilities learnt by chemical language models with the beam search algorithm as a model-intrinsic technique for automated molecule design and scoring. Prospective application of this method yielded novel inverse agonists of retinoic acid receptor-related orphan receptors (RORs). Each design was synthesizable in three reaction steps and presented low-micromolar to nanomolar potency towards RORγ. This model-intrinsic sampling technique eliminates the strict need for external compound scoring functions, thereby further extending the applicability of generative artificial intelligence to data-driven drug discovery.

Full text of DOI:10.1002/anie.202104405

Angewandte
Research Articles
Chemie
How to cite:
International Edition: doi.org/10.1002/anie.202104405
German Edition: doi.org/10.1002/ange.202104405
De novo Design
Beam Search for Automated Design and Scoring of Novel ROR
Ligands with Machine Intelligence**
Michael Moret+, Moritz Helmstꢀdter+, Francesca Grisoni, Gisbert Schneider,* and Daniel Merk*
Abstract: Chemical language models enable de novo drug
design without the requirement for explicit molecular con-
struction rules. While such models have been applied to
generate novel compounds with desired bioactivity, the actual
prioritization and selection of the most promising computa-
tional designs remains challenging. Herein, we leveraged the
probabilities learnt by chemical language models with the
beam search algorithm as a model-intrinsic technique for
automated molecule design and scoring. Prospective applica-
tion of this method yielded novel inverse agonists of retinoic
acid receptor-related orphan receptors (RORs). Each design
was synthesizable in three reaction steps and presented low-
micromolar to nanomolar potency towards RORg. This
model-intrinsic sampling technique eliminates the strict need
for external compound scoring functions, thereby further
extending the applicability of generative artificial intelligence
to data-driven drug discovery.
Figure 1. Molecule generation with a chemical language model (CLM)
and beam search sampling. a) Kekulꢀ structure of an example mole-
cule and corresponding SMILES string. b) CLM training. The CLM
Introduction
learns to predict the probability of each SMILES string character
Generative deep learning,[1,2] that is, a class of machine
(“token”) based on the previous tokens in the string. c) Beam search
decoding of width two (k=2): The design algorithm keeps track of the
learning models able to generate new data, can be applied to
computationally design pharmacologically active compounds
de novo.[3–5] Deep learning-based molecular design algo-
rithms can extract high-level molecular features from “raw”
molecular representations,[6–10] such as molecular graphs and
the Simplified Molecular Input Line Entry System (SMILES,
Figure 1a),[11] potentially allowing them to access unexplored
regions of the chemical space.[12] Previous studies showed that
chemical language models (CLMs),[13,14] in particular gener-
ative deep learning models trained on SMILES strings, can
generate novel molecules with experimentally validated
two most likely SMILES strings (highlighted in color). In this example,
the SMILES string generation proceeds from left to right.
bioactivity.[9,15,16] CLMs have shown the ability to learn
focused chemical features from small collections of template
molecules by means of transfer learning, that is, a method to
reuse previously learned knowledge on a new task for which
the available data is scarce.[15,17,18] Transfer learning is
performed in two steps. In the first step, a model is trained
[*] M. Moret,[+] Prof. Dr. F. Grisoni, Prof. Dr. G. Schneider
ETH Zurich, Department of Chemistry and Applied Biosciences
Vladimir-Prelog-Weg 4, 8093 Zurich (Switzerland)
E-mail: gisbert@ethz.ch
Prof. Dr. D. Merk
LMU Munich, Department of Pharmacy
Butenandtstrasse 7, 81377 Munich (Germany)
[+] These authors contributed equally to this work.
M. Helmstꢁdter,[+] Prof. Dr. D. Merk
Goethe University Frankfurt
[**] A previous version of this manuscript has been deposited on
a preprint server (http://doi.org/10.26434/chemrxiv.14153408.v1).
Institute of Pharmaceutical Chemistry
Max-von-Laue-Strasse 9, 60438 Frankfurt (Germany)
E-mail: merk@pharmchem.uni-frankfurt.de
Supporting information and the ORCID identification number(s) for
the author(s) of this article can be found under:
https://doi.org/10.1002/anie.202104405.
Prof. Dr. F. Grisoni
ꢂ 2021 The Authors. Angewandte Chemie International Edition
published by Wiley-VCH GmbH. This is an open access article under
the terms of the Creative Commons Attribution License, which
permits use, distribution and reproduction in any medium, provided
the original work is properly cited.
Eindhoven University of Technology
Institute for Complex Molecular Systems
Department of Biomedical Engineering
Groene Loper 7, 5612AZ Eindhoven (Netherlands)
Prof. Dr. G. Schneider
ETH Singapore SEC Ltd
1 CREATE Way, #06-01 CREATE Tower
Singapore 138602 (Singapore)
&&&&
ꢂ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 2 – 8
These are not the final page numbers!

Angewandte
Research Articles
Chemie
on a large amount of data that relate to the task to be Results and Discussion
performed (“pre-training”). In the case of CLMs, this is
usually done using large collections of molecules (e.g., in the
order of 200000 to 1000000[9,16,17]). Pre-training enables the
generative model to capture a) the SMILES “syntax” (i.e.,
how alphanumeric characters should be assembled to gen-
erate strings that correspond to valid molecules, Figure 1) and
b) the properties of the pre-training dataset, such as phys-
icochemical features and synthesizability of the molecules in
the dataset. In the second step, the pre-trained CLM is further
trained (“fine-tuned”) with a smaller set of task-specific
molecules.[13,19,20] During this transfer learning process, the
CLM is biased towards the chemical space of interest, that is,
molecules with desired biological and physicochemical prop-
erties. This ability to learn in a low-data regime (“few-shot”
learning[21,22]) renders CLMs particularly useful for applica-
tion to biological targets for which only few ligands are
known. The fully trained CLM can be used to generate new
molecules in the form of SMILES strings. Such data
generation is performed by predicting one character of
a SMILES string (“token”) at a time, based on all the
previous tokens. Importantly, this process does not require
handcrafted molecule design rules, as CLMs learn solely from
the SMILES strings used for training.
Previous prospective applications of CLMs for de novo
molecule generation used the so-called “temperature sam-
pling” to generate large virtual molecular libraries.[9,13,15]
Temperature sampling allows to sample new SMILES strings
by adding tokens to the (growing) string according to the
probabilities learned by the CLM, wherein the most likely
token at a given position will be sampled more often
(Figure 1b). However, the generated SMILES strings might
not always be “chemically meaningful” (invalid strings), or
they might not match the feature distribution of the training
data because of the random component of temperature
sampling. Therefore, additional methods are usually needed
to select the most promising designs from the virtual
molecular libraries, e.g., based on the similarity to known
bioactive molecules, external activity prediction, or reward
functions.[9,13,15,23] Here, we use the beam search algorithm as
a model-intrinsic alternative to temperature sampling. This
method enables the CLM to simultaneously generate and
prioritize the molecular designs in an automated fashion,
without employing additional selection methods.[24,25] Beam
search scoring was successfully validated in a prospective
application aiming to generate new retinoic acid-related
orphan receptor (ROR)[26] ligands from scratch.
Chemical Language Model and Beam Search Sampling for
De Novo Design
We explored the beam search algorithm[33] to generate
molecules from a CLM as a potential alternative to temper-
ature sampling combined with an external ranking method.
Given the probabilities learnt by a CLM, a vast number of
SMILES strings could in theory be sampled. As it is
computationally not feasible to sample all outputs, a heuristic
method such as beam search can be used to find the likely
outputs. Here, our underlying hypothesis was that the
probability for generating a certain SMILES string correlates
with the quality of the corresponding molecule regarding the
implicit design objective as represented in the fine-tuning set
(e.g., desired bioactivity, physicochemical properties). During
molecule generation by beam search sampling, the algorithm
progressively adds tokens to a SMILES string while keeping
track of the k most likely SMILES string(s). To add a new
token, the algorithm computes the conditional probability of
each possible token given the tokens in the existing string and
defines the k most likely tokens to extend the string (Fig-
ure 1c). The set of k most likely selections is based on
a scoring function (“beam search score”), which is computed
as the product of the probabilities of each token (Figure 1c).
This process is repeated until the SMILES string is completed
(i.e., the “end-of-string” token is added) or a predefined
maximal string length is reached. Thus, beam search can be
used to generate highly probable molecules, as computed by
(i) the underlying model and (ii) the beam search score. The
beam search score allows to rank the de novo designs
according to the probability of their SMILES tokens.
As a framework to probe beam search sampling, we
employed a recently published CLM based on a recurrent
neural network with long short-term memory cells (LSTM),
which are suited for sequence modeling.[34] The CLM was
trained with the SMILES strings of 365063 molecules from
ChEMBL[35] to iteratively predict the next token of each
SMILES string given the preceding tokens (Figure 1b). The
training procedure was carried out over ten epochs, meaning
that each molecule used for training was seen by the CLM ten
times. This pre-trained CLM was then fine-tuned using sets of
known ROR ligands (Figure S1, Table S1), to obtain a bias
towards the design objective, namely the generation of new
molecules with bioactivity on RORs, by transfer learning.
Open-source code for the CLM and the beam search
algorithm, and the data used in this study are available at
https://github.com/ETHmodlab/
RORs were chosen as molecular targets because these
receptor proteins are an attractive but not extensively studied
family of potential drug targets. They constitute a family of
ligand-activated transcription factors that mainly act as
monomers and are involved in the circadian control of energy
homeostasis[27,28] and immune system regulation,[29,30] among
other functions. RORs hold promising pharmacological
potential for various indications, specifically for autoimmune
diseases.[29,30] No ROR ligand has reached drug approval to
date, partially owing to compound-related issues such as poor
aqueous solubility, lack of selectivity, and clinical safety
concerns.[29,31,32]
molecular design with beam search.
Application of Beam Search Sampling to Designing Inverse
RORg Agonists
For prospective evaluation, we applied the beam search to
the design of natural product-inspired RORg ligands. Learn-
ing from natural products as a traditional source of inspiration
for drug discovery[36,37] may hold several advantages over
Angew. Chem. Int. Ed. 2021, 60, 2 – 8
ꢂ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
www.angewandte.org
&&&&
These are not the final page numbers!

Angewandte
Research Articles
Chemie
learning from purely synthetic molecules, because of the
overall higher structural diversity, greater three-dimension-
ality, and often superior selectivity of bioactive natural
products.[38,39] We aimed to obtain de novo designs possessing
three properties: (i) natural product-inspired chemical struc-
ture, (ii) synthesizability, and (iii) bioactivity on RORg. Aim-
ing to fulfil all three objectives during transfer learning, the
previously pre-trained CLM on bioactive molecules from
ChEMBL[17] was fine-tuned on one synthetic and four natural
product RORg modulators described in literature[30] (Fig-
ure S1). From the fine-tuned model, beam search sampling
was started after the fifth epoch of fine-tuning, to ensure that
the CLM had sufficiently captured the molecular features of
the small fine-tuning set.
All valid SMILES strings generated between epochs 5 and
16 (last fine-tuning epoch) were ranked by beam search
scoring. The top five designs according to the beam search
score (Figure 2a) were deemed synthetically inaccessible by
medicinal chemists. This was further highlighted by the
predictions of a machine learning algorithm for retrosynthetic
analysis (IBM RXN)[40] which did not find a synthetic route
for any of these molecules. Thus, while the CLM captured
natural product likeness, the model failed to meet the generic
design criterion of synthesizability. These findings point to
a benefit of beam search sampling in revealing the most likely
CLM molecules to assess the success of fine-tuning in terms of
the design objectives.
the model towards natural-product-likeness. Again, valid
SMILES strings generated by beam search sampling between
epochs 5 and 16 of the (second) fine-tuning step were
considered. With this second approach, the top 5 sampled
molecules (Figure 2b) were synthetically accessible according
to IBM RXN,[40] which could propose a synthetic route for
each of them. Importantly, the computer-generated molecules
possess certain natural product characteristics (Figure 3,
Table S2), as indicated by a high fraction of sp3-hybridized
carbon atoms (Fsp3). The top five designs have Fsp3 values
ranging from 50% to 75%. These values are comparable to
those computed for the MEGx natural product library
(Analyticon Discovery GmbH, rel. 09-01-2018), and exceed
the average Fsp3 value of the ChEMBL molecules used for
pre-training (51 ꢀ 30% and 33 ꢀ 20%, respectively). These
results suggested that the two-step fine-tuning procedure
complied with the design objectives and the implemented
two-step approach was chosen for prospective application.
We then compared the beam search designs obtained with
the chosen computational strategy to known RORg modu-
lators and to the fine-tuning molecules (Figure 3a,b). Despite
favoring only some of the most likely tokens while generating
new SMILES strings, and examining only a limited set of
possibilities, the beam search sampling still allowed to explore
the chemical space beyond the regions that are populated by
the fine-tuning compounds (Figure 3a). Compared to the
inverse RORg agonists annotated in ChEMBL (IC50 < 1 mm,
Figure 2d), the beam search designs are structurally more
diverse in terms of substructure fragments, as represented by
Morgan fingerprints.[42] Still, the designs possess a certain
degree of similarity to the known active molecules in terms of
their three-dimensional shape and partial charge distribution
(as represented by the Weighted Holistic Atom Localization
and Entity Shape [WHALES] descriptors[43,44]). Apparently,
the CLM, in addition to learning the SMILES “syntax”, also
Aiming to improve upon these results, we performed
a second experiment in which we applied a two-step fine-
tuning strategy. First, the pre-trained model was fine-tuned
for 20 epochs on 255 synthetic RORg ligands from the
US patent subset of the Protein Data Bank[41] (255 molecules,
Table S1) to capture both bioactivity and synthesizability.
Then, the model was fine-tuned with four natural product
RORg modulators[30] (Figure S1) for 16 epochs, aiming to bias
Figure 2. Top-ranked designs obtained by beam search sampling. a) Single fine-tuning, b) double fine-tuning. Ranks are based on the beam
search score of the molecular designs. For the top-ranked molecules from the double fine-tuning experiment, the similarity values refer to the
Tanimoto similarity computed on Morgan fingerprints (length=1024, 2-bond radius) to the closest known active molecule annotated in ChEMBL
with an IC50 value for RORg (structures are shown in Figure S2).
&&&&
www.angewandte.org
ꢂ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 2 – 8
These are not the final page numbers!

Angewandte
Research Articles
Chemie
design ranked 6th. Molecules ranked 10th and 13th resembled
compound 2. Hence, we additionally chose compound 3 of
this abundant chemotype from rank 13 for prospective
validation. Compounds 1–3 were synthesized according to
Scheme 1.
For preparation of 1, (4-chlorophenyl)piperazine (4) was
treated with 4-bromobutyl acetate (5) to obtain the ester-
protected intermediate 6 which after alkaline ester hydrolysis
to 7 was suitable for Mitsunobu reaction with 8-azaspiro-
[4.5]decane-7,9-dione (8) to obtain the top-ranked computa-
tional design 1. Preparation of 2 started from 4-bromo-2-
fluorobenzaldehyde (9) which was reacted with amine 10 to
obtain 11 by reductive amination followed by sulfonamide
coupling with 12 to yield 13. Eventually, the 4-trifluorome-
thylpiperidine substituent was introduced to 13 under Buch-
wald–Hartwig conditions with 14 yielding compound 2. The
structurally related design 3 was prepared via a different route
starting from a nucleophilic aromatic substitution of 4-
fluorobenzaldehyde (15) with 4-trifluoromethylpiperidine
(14) to 16. The nucleophilic aromatic substitution provided
substantially higher yield (see Scheme 1) than the Buchwald–
Hartwig reaction but could not be employed in the synthesis
of 2 because of the potential formation of regioisomers.
Reductive amination of 16 with cyclobutaneamine (10) to 17,
followed by sulfonamide formation with phenylmethanesul-
fonyl chloride (12), afforded the computationally designed
compound 3.
In vitro characterization of compounds 1, 2, and 3 in Gal4-
ROR hybrid reporter gene assays confirmed inverse RORg
agonism with micromolar to sub-micromolar IC50 values
(Table 1). The top-ranked compound 1 counteracted RORg
activity with an IC50 value of 4.6 mm. It was additionally active
on RORa and RORb, but precise IC50 values could not be
determined due to cytotoxicity. Compounds 2 and 3 blocked
RORg activity with IC50 values of 0.37 mm (2) and 0.68 mm (3),
respectively. In addition to being inverse RORg agonists, all
three synthesized designs revealed pronounced preference for
the RORg subtype, with compounds 2 and 3 possessing more
than tenfold higher potency on RORg compared to the
related RORa and RORb isoforms. These results show that
the CLM with beam search sampling conserved the bioactiv-
ity of the training molecules in the computational designs.
Figure 3. Characteristics of designs from the CLM with double fine-
tuning. a) Stochastic neighbor embedding (t-SNE)[45] projection of the
compound sets based on Morgan fragment fingerprints
(length=1024, 2-bond radius, Tanimoto similarity). The location of the
two-fine tuning sets, the RORg modulators annotated in ChEMBL
(IC50 <1 mm, 1091 compounds), and the beam search designs are
shown. b) Comparison of the sampled molecular designs with known
RORg modulators (IC50 <1 mm) in terms of Morgan fragment finger-
prints (“Morgan”) and three-dimensional shape and electrostatics
descriptors (WHALES). The pairwise distance distribution among
known RORg modulators contained in ChEMBL is shown as a refer-
ence. For Morgan fingerprints, the Tanimoto distance is shown; for
WHALES the range-scaled Euclidean distance is shown. “Beam (15)”
and “Beam (5)” indicate the top 15 and top 5 designs, respectively.
Boxplots indicate 25th, 50th, and 75th percentiles (lines), mean values
(circle), and outlier boundaries (whiskers, 1.5ꢃ interquartile range).
learned certain “semantic” ligand features that are relevant
for binding to macromolecules, such as their molecular shape Conclusion
and partial charge patterns.
Herein, Beam search sampling from CLMs was applied to
generating new molecules with desired bioactivity on the
ligand-activated transcription factor RORg. The algorithm
automatically generated and scored the designs, without the
need of additional prioritization rules. Prospective experi-
mental validation yielded three novel, potent inverse agonists
of the nuclear receptor with various degrees of similarity to
known RORg modulators (ranging from 0.28 to 0.71, as
captured by Tanimoto similarity on Morgan fingerprints).
Apparently, the beam search approach coupled with a CLM
conserves structural features necessary for the desired
bioactivity but still generates structurally diverse compounds
in terms of fragments. This observation corroborates beam
Prospective Experimental Validation
Three beam search designs were synthesized and charac-
terized in vitro. We selected them based on their beam search
score. From the five most likely designs (Figure 2b), we
selected molecules 1 and 2, which were ranked first and third.
Compound 2 showed the highest Tanimoto similarity (Mor-
gan fingerprints) to a known RORg modulator (Figure 2b).
The scaffolds of both compounds were also prominent among
the beam search designs not included in the top 5, suggesting
a structural preference. The scaffold of 1 also appeared in the
Angew. Chem. Int. Ed. 2021, 60, 2 – 8
ꢂ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
www.angewandte.org
&&&&
These are not the final page numbers!

Angewandte
Research Articles
Chemie
Scheme 1. Synthesis of the CLM designs 1, 2, and 3. Reagents and conditions: a) DMF, 4-DMAP, 608C, 16 h, 48%; b) KOH, H2O/THF/MeOH,
microwave irradiation, 1008C, 30 min, 98%; c) DIAD, PPh3, THF, 08C!r.t., 16 h, 42%; d) NaB(OAc)3H, HOAc, DCE, r.t., 50 h, 73%; e) 4-DMAP,
pyridine, CH2Cl2, reflux, 16 h, 37%; f) Pd2(dba)3, xantphos, Cs2CO3, 1,4-dioxane, reflux, 16 h, 18%; g) K2CO3, DMSO, reflux, 48 h, 82%.
Table 1: Activity of de novo designs 1, 2, and 3 on RORs in Gal4 hybrid
compliance of the CLM designs with the design objectives
reporter gene assays. Data are reported as meanꢀS.E.M., nꢁ4.
and to assess the success of fine-tuning. This is in contrast to
standard temperature sampling, which might lead chemists to
IC50 [mm]
RORb
Structure and ID
RORa
RORg
consider designs that are not likely according to the model.
Secondly, beam search sampling could potentially overcome
the need for external compound prioritization. It should be
noted, however, that the number of designs that can be
sampled by beam search is limited compared to temperature
sampling, which can virtually generate an infinite number of
chemical structures. The two techniques complement each
other, and both offer characteristic advantages. The desired
application should guide the choice of either strategy. If
corroborated in future prospective studies, beam search
sampling may help to further the applicability of CLMs for
de novo molecular design in medicinal chemistry.
>10
>10
4.6ꢀ0.5
23ꢀ3
10ꢀ1
22ꢀ1
0.37ꢀ0.05
0.68ꢀ0.07
7.6ꢀ0.5
Acknowledgements
This research was supported by the Swiss National Science
Foundation (grant no. 205321_182176 to G.S.), the RE-
THINK initiative at ETH Zurich, and the Novartis For-
schungsstiftung (FreeNovation grant “AI in Drug Discovery”
to G.S.). Open access funding enabled and organized by
Projekt DEAL.
search sampling as a technique for the de novo design of
bioactive molecules by a CLM. The computational and
experimental results suggest two attractive properties of the
beam search algorithm. Firstly, by searching for the most
likely molecules a CLM can generate, the beam search
algorithm probes the suitability of a CLM for the given task.
Evaluation of the resulting designs allows to check the
&&&&
www.angewandte.org
ꢂ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 2 – 8
These are not the final page numbers!

Angewandte
Research Articles
Chemie
[23] X. Yang, J. Zhang, K. Yoshizoe, K. Terayama, K. Tsuda, Sci.
Conflict of Interest
Technol. Adv. Mater. 2017, 18, 972 – 976.
[24] P. Schwaller, T. Laino, T. Gaudin, P. Bolgar, C. A. Hunter, C.
Bekas, A. A. Lee, ACS Cent. Sci. 2019, 5, 1572 – 1583.
[25] D. Grechishnikova, Sci. Rep. 2021, 11, 321.
[26] G. Benoit, A. Cooney, V. Giguere, H. Ingraham, M. Lazar, G.
Muscat, T. Perlmann, J. P. Renaud, J. Schwabe, F. Sladek, M. J.
Tsai, V. Laudet, Pharmacol. Rev. 2006, 58, 798 – 836.
[27] D. P. Marciano, M. R. Chang, C. A. Corzo, D. Goswami, V. Q.
Lam, B. D. Pascal, P. R. Griffin, Cell Metab. 2014, 19, 193 – 208.
[28] Y. Hoon Kim, M. A. Lazar, Endocr. Rev. 2020, 41, 707 – 732.
[29] V. B. Pandya, S. Kumar, Sachchidanand, R. Sharma, R. C. Desai,
J. Med. Chem. 2018, 61, 10976 – 10995.
G.S. declares a potential financial conflict of interest as
a founder of inSili.com GmbH, Zurich, and in his role as
consultant to the pharmaceutical industry.
Keywords: de novo design · deep learning · drug discovery ·
neural network · nuclear receptor
[1] J. Schmidhuber, Neural Networks 2015, 61, 85 – 117.
[2] Y. Lecun, Y. Bengio, G. Hinton, Nature 2015, 521, 436 – 444.
[3] H. Chen, O. Engkvist, Y. Wang, M. Olivecrona, T. Blaschke,
Drug Discov. Today 2018, 23, 1241 – 1250.
[30] L. A. Solt, T. P. Burris, Trends Endocrinol. Metab. 2012, 23, 619 –
627.
[31] S. Asimus, R. Palmꢃr, M. Albayaty, H. Forsman, C. Lundin, M.
Olsson, R. Pehrson, J. Mo, M. Russell, S. Carlert, D. Close, D.
Keeling, Br. J. Clin. Pharmacol. 2020, 86, 1398 – 1405.
[32] D. J. Kojetin, T. P. Burris, Nat. Rev. Drug Discovery 2014, 13,
197 – 216.
[4] W. P. Walters, R. Barzilay, Acc. Chem. Res. 2021, 54, 263 – 270.
[5] B. Sanchez-Lengeling, A. Aspuru-Guzik, Science 2018, 361,
360 – 365.
[6] R. Gꢀmez-Bombarelli, J. N. Wei, D. Duvenaud, J. M. Hernꢁn-
dez-Lobato, B. Sꢁnchez-Lengeling, D. Sheberla, J. Aguilera-
Iparraguirre, T. D. Hirzel, R. P. Adams, A. Aspuru-Guzik, ACS
Cent. Sci. 2018, 4, 268 – 276.
[33] B. T. Lowerre, PhD Thesis Carnegie Mellon Univ. Pittsburgh,
1976.
[34] S. Hochreiter, J. Schmidhuber, Neural Comput. 1997, 9, 1735 –
1780.
[35] A. P. Bento, A. Gaulton, A. Hersey, L. J. Bellis, J. Chambers, M.
Davies, F. A. Krꢂger, Y. Light, L. Mak, S. McGlinchey, M.
Nowotka, G. Papadatos, R. Santos, J. P. Overington, Nucleic
Acids Res. 2014, 42, D1083 – D1090.
[36] D. J. Newman, G. M. Cragg, J. Nat. Prod. 2020, 83, 770 – 803.
[37] T. Rodrigues, D. Reker, P. Schneider, G. Schneider, Nat. Chem.
2016, 8, 531 – 541.
[7] N. de Cao, T. Kipf, arXiv 2018, https://arxiv.org/abs/1805.11973.
[8] A. Gupta, A. T. Mꢂller, B. J. H. Huisman, J. A. Fuchs, P.
Schneider, G. Schneider, Mol. Inf. 2018, 37, 1700111.
[9] D. Merk, L. Friedrich, F. Grisoni, G. Schneider, Mol. Inf. 2018,
37, 1700153.
[10] J. Bradshaw, B. Paige, M. J. Kusner, M. H. S. Segler, J. M.
Hernꢁndez-Lobato, arXiv 2020, https://arxiv.org/abs/2012.11522.
[11] D. Weininger, J. Chem. Inf. Comput. Sci. 1988, 28, 31 – 36.
[12] M. A. Skinnider, R. G. Stacey, D. S. Wishart, L. J. Foster,
[38] P. Ertl, A. Schuffenhauer, in Prog. Drug Res., Birkhꢄuser, Basel,
2008, pp. 217 – 235.
ChemRxiv
13638347.V1.
2021,
https://doi.org/10.26434/CHEMRXIV.
[39] P. Ertl, S. Roggo, A. Schuffenhauer, J. Chem. Inf. Model. 2008,
48, 68 – 74.
[13] M. H. S. Segler, T. Kogej, C. Tyrchan, M. P. Waller, ACS Cent.
Sci. 2018, 4, 120 – 131.
[40] P. Schwaller, R. Petraglia, V. Zullo, V. H. Nair, R. A. Haeusel-
mann, R. Pisoni, C. Bekas, A. Iuliano, T. Laino, Chem. Sci. 2020,
11, 3316 – 3325.
[41] H. M. Berman, T. Battistuz, T. N. Bhat, W. F. Bluhm, P. E.
Bourne, K. Burkhardt, Z. Feng, G. L. Gilliland, L. Iype, S. Jain, P.
Fagan, J. Marvin, D. Padilla, V. Ravichandran, B. Schneider, N.
Thanki, H. Weissig, J. D. Westbrook, C. Zardecki, Acta Crys-
tallogr. Sect. D 2002, 58, 899 – 907.
[42] H. L. Morgan, J. Chem. Doc. 1965, 5, 107 – 113.
[43] F. Grisoni, D. Merk, V. Consonni, J. A. Hiss, S. G. Tagliabue, R.
Todeschini, G. Schneider, Commun. Chem. 2018, 1, 44.
[44] F. Grisoni, G. Schneider, Methods Mol. Biol. 2021, 2266, 11 – 35.
[45] L. van der Maaten, G. Hinton, J. Mach. Learn. Res. 2008, 9,
2579 – 2605.
[14] W. Yuan, D. Jiang, D. K. Nambiar, L. P. Liew, M. P. Hay, J.
Bloomstein, P. Lu, B. Turner, Q. T. Le, R. Tibshirani, P. Khatri,
M. G. Moloney, A. C. Koong, J. Chem. Inf. Model. 2017, 57, 875 –
882.
[15] D. Merk, F. Grisoni, L. Friedrich, G. Schneider, Commun. Chem.
2018, 1, 68.
[16] Y. Yang, R. Zhang, Z. Li, L. Mei, S. Wan, H. Ding, Z. Chen, J.
Xing, H. Feng, J. Han, H. Jiang, M. Zheng, C. Luo, B. Zhou, J.
Med. Chem. 2020, 63, 1337 – 1360.
[17] M. Moret, L. Friedrich, F. Grisoni, D. Merk, G. Schneider, Nat.
Mach. Intell. 2020, 2, 171 – 180.
[18] M. Awale, F. Sirockin, N. Stiefl, J. L. Reymond, J. Chem. Inf.
Model. 2019, 59, 1347 – 1356.
[19] J. Yosinski, J. Clune, Y. Bengio, H. Lipson, Adv. Neural Inf.
Process. Syst. 2014, 27, 3320 – 3328.
[20] M. Peters, S. Ruder, N. A. Smith, arXiv 2019, https://arxiv.org/
abs/1903.05987.
[21] H. Altae-Tran, B. Ramsundar, A. S. Pappu, V. Pande, ACS Cent.
Sci. 2017, 3, 283 – 293.
[22] Y. Wang, Q. Yao, J. Kwok, L. M. Ni, arXiv 2019, https://arxiv.org/
abs/1904.05046.
Manuscript received: March 30, 2021
Revised manuscript received: June 2, 2021
Accepted manuscript online: June 24, 2021
Version of record online: && &&
,
&&&&
Angew. Chem. Int. Ed. 2021, 60, 2 – 8
ꢂ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
www.angewandte.org
&&&&
These are not the final page numbers!

Angewandte
Research Articles
Chemie
Research Articles
De novo Design
M. Moret, M. Helmstꢁdter, F. Grisoni,
G. Schneider,* D. Merk*
&&&& — &&&&
Beam Search for Automated Design and
Scoring of Novel ROR Ligands with
Machine Intelligence
The beam search algorithm was
novel ligands of this nuclear receptor with
the intended bioactivity. Beam search
sampling overcomes the need for external
scoring methods and extends the applic-
ability of machine learning-driven molec-
ular design.
employed for automated molecule design
and scoring from a chemical language
model (CLM). Prospective application of
this model-intrinsic technique with a CLM
trained on inverse RORg agonists yielded
&&&&
www.angewandte.org
ꢂ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 2 – 8
These are not the final page numbers!