Improved Selection of Rare Reactions in Template-Based Retrosynthesis Predictions

Computer Assisted Synthesis Planning (CASP) has gained considerable interest as of late. Herein we investigate a template-based retrosynthetic planning tool, trained on a variety of datasets consisting of up to 17.5 million reactions. We demonstrate that models trained on datasets such as internal Electronic Laboratory Notebooks (ELN), and the publicly available United States Patent Office (USPTO) extracts, are sufficient for the prediction of full synthetic routes to compounds of interest in medicinal chemistry. As such we have assessed the models on 1,731 compounds from 41 virtual libraries for which experimental results were known. Furthermore, we show that accuracy is a misleading metric for assessment of the ‘filter network’, and propose that the number of successfully applied templates, in conjunction with the overall ability to generate full synthetic routes be examined instead. To this end we found that the specificity of the templates comes at the cost of generalizability, and overall model performance. This is supplemented by a comparison of the underlying datasets and their corresponding models.

Download Full-text

CompRet: A Comprehensive Recommendation Framework for Chemical Synthesis Planning with Algorithmic Enumeration

10.26434/chemrxiv.12295976 ◽

2020 ◽

Author(s):

Ryosuke Shibukawa ◽

Shoichi Ishida ◽

Kazuki Yoshizoe ◽

Kunihiro Wasa ◽

Kiyosei Takasu ◽

...

Keyword(s):

Chemical Synthesis ◽

Chemical Reaction ◽

Reaction Network ◽

Exact Algorithm ◽

Chemical Reaction Network ◽

Computer Assisted ◽

Enumeration Algorithm ◽

Synthesis Planning ◽

Synthetic Routes ◽

Alternative Routes

In computer-assisted synthesis planning (CASP) programs, providing chemical synthetic routes as many as possible is essential for considering optimal and alternative routes in a chemical reaction network. As the majority of CASP programs have been designed to provide one or a few optimal routes, it is likely that desired one will not be included. To avoid this, an exact algorithm that lists possible synthetic routes from the chemical reaction network is required, alongside a recommendation of synthetic routes that meet specified criteria based on chemist's objectives. Herein, we propose a chemical-reaction-network-based synthetic route recommendation framework called "CompRet" with a mathematically guaranteed enumeration algorithm. In a preliminary experiment, CompRet was shown to successfully provide alternative routes for a known antihistaminic drug, cetirizine. CompRet is expected to promote desirable enumeration-based chemical synthesis searches and aid the development of an interactive CASP framework for chemists.

Download Full-text

CompRet: A Comprehensive Recommendation Framework for Chemical Synthesis Planning with Algorithmic Enumeration

10.26434/chemrxiv.12295976.v1 ◽

2020 ◽

Author(s):

Ryosuke Shibukawa ◽

Shoichi Ishida ◽

Kazuki Yoshizoe ◽

Kunihiro Wasa ◽

Kiyosei Takasu ◽

...

Keyword(s):

Chemical Synthesis ◽

Chemical Reaction ◽

Reaction Network ◽

Exact Algorithm ◽

Chemical Reaction Network ◽

Computer Assisted ◽

Enumeration Algorithm ◽

Synthesis Planning ◽

Synthetic Routes ◽

Alternative Routes

In computer-assisted synthesis planning (CASP) programs, providing chemical synthetic routes as many as possible is essential for considering optimal and alternative routes in a chemical reaction network. As the majority of CASP programs have been designed to provide one or a few optimal routes, it is likely that desired one will not be included. To avoid this, an exact algorithm that lists possible synthetic routes from the chemical reaction network is required, alongside a recommendation of synthetic routes that meet specified criteria based on chemist's objectives. Herein, we propose a chemical-reaction-network-based synthetic route recommendation framework called "CompRet" with a mathematically guaranteed enumeration algorithm. In a preliminary experiment, CompRet was shown to successfully provide alternative routes for a known antihistaminic drug, cetirizine. CompRet is expected to promote desirable enumeration-based chemical synthesis searches and aid the development of an interactive CASP framework for chemists.

Download Full-text

Datasets and Their Influence on the Development of Computer Assisted Synthesis Planning Tools in the Pharmaceutical Domain

10.26434/chemrxiv.9897692.v1 ◽

2019 ◽

Author(s):

Amol Thakkar ◽

Thierry Kogej ◽

Jean-Louis Reymond ◽

Ola Engkvist ◽

Esben Jannik Bjerrum

Keyword(s):

Model Performance ◽

Computer Assisted ◽

Patent Office ◽

United States Patent ◽

Planning Tool ◽

Virtual Libraries ◽

Synthesis Planning ◽

Synthetic Routes ◽

The Cost ◽

States Patent

Computer Assisted Synthesis Planning (CASP) has gained considerable interest as of late. Herein we investigate a template-based retrosynthetic planning tool, trained on a variety of datasets consisting of up to 17.5 million reactions. We demonstrate that models trained on datasets such as internal Electronic Laboratory Notebooks (ELN), and the publicly available United States Patent Office (USPTO) extracts, are sufficient for the prediction of full synthetic routes to compounds of interest in medicinal chemistry. As such we have assessed the models on 1,731 compounds from 41 virtual libraries for which experimental results were known. Furthermore, we show that accuracy is a misleading metric for assessment of the ‘filter network’, and propose that the number of successfully applied templates, in conjunction with the overall ability to generate full synthetic routes be examined instead. To this end we found that the specificity of the templates comes at the cost of generalizability, and overall model performance. This is supplemented by a comparison of the underlying datasets and their corresponding models.

Download Full-text

Predicting Retrosynthetic Reaction using Self-Corrected Transformer Neural Networks

10.26434/chemrxiv.8427776.v1 ◽

2019 ◽

Author(s):

Shuangjia Zheng ◽

Jiahua Rao ◽

Zhongyue Zhang ◽

Jun Xu ◽

Yuedong Yang

Keyword(s):

Neural Network ◽

Network Architecture ◽

State Of The Art ◽

Neural Network Architecture ◽

Training Set ◽

Computer Aided ◽

Synthesis Planning ◽

Target Molecules ◽

Template Free ◽

Synthetic Routes

<a>Synthesis planning is the process of recursively decomposing target molecules into available precursors. Computer-aided retrosynthesis can potentially assist chemists in designing synthetic routes, but at present it is cumbersome and provides results of dissatisfactory quality. In this study, we develop a template-free self-corrected retrosynthesis predictor (SCROP) to perform a retrosynthesis prediction task trained by using the Transformer neural network architecture. In the method, the retrosynthesis planning is converted as a machine translation problem between molecular linear notations of reactants and the products. Coupled with a neural network-based syntax corrector, our method achieves an accuracy of 59.0% on a standard benchmark dataset, which increases >21% over other deep learning methods, and >6% over template-based methods. More importantly, our method shows an accuracy 1.7 times higher than other state-of-the-art methods for compounds not appearing in the training set.</a>

Download Full-text

Predicting Retrosynthetic Reaction using Self-Corrected Transformer Neural Networks

10.26434/chemrxiv.8427776 ◽

2019 ◽

Author(s):

Shuangjia Zheng ◽

Jiahua Rao ◽

Zhongyue Zhang ◽

Jun Xu ◽

Yuedong Yang

Keyword(s):

Neural Network ◽

Network Architecture ◽

State Of The Art ◽

Neural Network Architecture ◽

Training Set ◽

Computer Aided ◽

Synthesis Planning ◽

Target Molecules ◽

Template Free ◽

Synthetic Routes

<a>Synthesis planning is the process of recursively decomposing target molecules into available precursors. Computer-aided retrosynthesis can potentially assist chemists in designing synthetic routes, but at present it is cumbersome and provides results of dissatisfactory quality. In this study, we develop a template-free self-corrected retrosynthesis predictor (SCROP) to perform a retrosynthesis prediction task trained by using the Transformer neural network architecture. In the method, the retrosynthesis planning is converted as a machine translation problem between molecular linear notations of reactants and the products. Coupled with a neural network-based syntax corrector, our method achieves an accuracy of 59.0% on a standard benchmark dataset, which increases >21% over other deep learning methods, and >6% over template-based methods. More importantly, our method shows an accuracy 1.7 times higher than other state-of-the-art methods for compounds not appearing in the training set.</a>

Download Full-text

Current Approaches and Tools Used in Drug Development against Parkinson’s Disease

Biomolecules ◽

10.3390/biom11060897 ◽

2021 ◽

Vol 11 (6) ◽

pp. 897

Author(s):

Oliwia Koszła ◽

Piotr Stępnicki ◽

Agata Zięba ◽

Angelika Grudzińska ◽

Dariusz Matosiuk ◽

...

Keyword(s):

Parkinson’S Disease ◽

Parkinson's Disease ◽

Neurodegenerative Disorder ◽

Treatment Options ◽

New Drugs ◽

Computer Assisted ◽

Therapeutic Effects ◽

New Compounds ◽

The Brain ◽

Selection Of

Parkinson’s disease is a progressive neurodegenerative disorder characterized by the death of nerve cells in the substantia nigra of the brain. The treatment options for this disease are very limited as currently the treatment is mainly symptomatic, and the available drugs are not able to completely stop the progression of the disease but only to slow it down. There is still a need to search for new compounds with the most optimal pharmacological profile that would stop the rapidly progressing disease. An increasing understanding of Parkinson’s pathogenesis and the discovery of new molecular targets pave the way to develop new therapeutic agents. The use and selection of appropriate cell and animal models that better reflect pathogenic changes in the brain is a key aspect of the research. In addition, computer-assisted drug design methods are a promising approach to developing effective compounds with potential therapeutic effects. In light of the above, in this review, we present current approaches for developing new drugs for Parkinson’s disease.

Download Full-text

Review: Studies on the Synthesis of Quinolone Derivatives with Their Antibacterial Activity (Part 1)

Current Organic Chemistry ◽

10.2174/1385272824999200427082108 ◽

2020 ◽

Vol 24 (8) ◽

pp. 817-854

Author(s):

Anil Kumar ◽

Nishtha Saxena ◽

Arti Mehrotra ◽

Nivedita Srivastava

Keyword(s):

Antibacterial Activity ◽

Antibacterial Properties ◽

Structure Activity Relationship ◽

New Drugs ◽

Activity Relationship ◽

Structure Activity ◽

Medicinal Properties ◽

Synthetic Routes ◽

Quinolone Derivatives

Quinolone derivatives have attracted considerable attention due to their medicinal properties. This review covers many synthetic routes of quinolones preparation with their antibacterial properties. Detailed study with structure-activity relationship among quinolone derivatives will be helpful in designing new drugs in this field.

Download Full-text

Design of Inhibitors for Glyceraldehyde-3-phosphate Dehydrogenase (GAPDH) Enzyme of Leishmania mexicana

Medicinal Chemistry ◽

10.2174/1573406415666190712111139 ◽

2020 ◽

Vol 16 (6) ◽

pp. 784-795

Author(s):

Krisnna M.A. Alves ◽

Fábio José Bonfim Cardoso ◽

Kathia M. Honorio ◽

Fábio A. de Molfetta

Keyword(s):

Molecular Dynamics ◽

Molecular Dynamics Simulations ◽

Binding Site ◽

Point Of View ◽

New Drugs ◽

Leishmania Mexicana ◽

Receptor Complexes ◽

Starting Point ◽

Dynamics Simulations ◽

Selection Of

Background:: Leishmaniosis is a neglected tropical disease and glyceraldehyde 3- phosphate dehydrogenase (GAPDH) is a key enzyme in the design of new drugs to fight this disease. Objective:: The present study aimed to evaluate potential inhibitors of GAPDH enzyme found in Leishmania mexicana (L. mexicana). Methods: A search for novel antileishmanial molecules was carried out based on similarities from the pharmacophoric point of view related to the binding site of the crystallographic enzyme using the ZINCPharmer server. The molecules selected in this screening were subjected to molecular docking and molecular dynamics simulations. Results:: Consensual analysis of the docking energy values was performed, resulting in the selection of ten compounds. These ligand-receptor complexes were visually inspected in order to analyze the main interactions and subjected to toxicophoric evaluation, culminating in the selection of three compounds, which were subsequently submitted to molecular dynamics simulations. The docking results showed that the selected compounds interacted with GAPDH from L. mexicana, especially by hydrogen bonds with Cys166, Arg249, His194, Thr167, and Thr226. From the results obtained from molecular dynamics, it was observed that one of the loop regions, corresponding to the residues 195-222, can be related to the fitting of the substrate at the binding site, assisting in the positioning and the molecular recognition via residues responsible for the catalytic activity. Conclusion:: he use of molecular modeling techniques enabled the identification of promising compounds as inhibitors of the GAPDH enzyme from L. mexicana, and the results obtained here can serve as a starting point to design new and more effective compounds than those currently available.

Download Full-text

Generating stable molecules using imitation and reinforcement learning

Machine Learning: Science and Technology ◽

10.1088/2632-2153/ac3eb4 ◽

2021 ◽

Author(s):

Søren Ager Meldgaard ◽

Jonas Köhler ◽

Henrik Lund Mortensen ◽

Mads-Peter Verner Christiansen ◽

Frank Noé ◽

...

Keyword(s):

Reinforcement Learning ◽

Chemical Space ◽

Training Data ◽

Graph Representation ◽

Imitation Learning ◽

Training Set ◽

Machine Learning Methods ◽

Multiple Copies ◽

The Stability ◽

3D Information

Abstract Chemical space is routinely explored by machine learning methods to discover interesting molecules, before time-consuming experimental synthesizing is attempted. However, these methods often rely on a graph representation, ignoring 3D information necessary for determining the stability of the molecules. We propose a reinforcement learning approach for generating molecules in cartesian coordinates allowing for quantum chemical prediction of the stability. To improve sample-efficiency we learn basic chemical rules from imitation learning on the GDB-11 database to create an initial model applicable for all stoichiometries. We then deploy multiple copies of the model conditioned on a specific stoichiometry in a reinforcement learning setting. The models correctly identify low energy molecules in the database and produce novel isomers not found in the training set. Finally, we apply the model to larger molecules to show how reinforcement learning further refines the imitation learning model in domains far from the training data.

Download Full-text