Reinforcement Learning for Bioretrosynthesis

Koch, Mathilde; Duigou, Thomas; Faulon, Jean-Loup

doi:10.1021/acssynbio.9b00447

Cited by 107 publications

(139 citation statements)

References 61 publications

Supporting

Mentioning

117

Contrasting

Unclassified

Order By: Relevance

“…Identification of these biosynthetic routes is accomplished by biochemically “walking back” from the target to precursor metabolites that are produced by, or fed to, the host organism. This procedure is called retrobiosynthesis and is implemented in a range of tools such as BNICE.ch 16,17 , GEM-Path 18 , NovoPathFinder 19 , NovoStoic 20 , ReactPRED 21 , RetroPath 22,23 , and Transform-MinER 24 . Retrobiosynthetic methods rely on the concept of generalized enzymatic reaction rules .…”

Section: Introductionmentioning

confidence: 99%

ATLASx: a computational map for the exploration of biochemical space

MohammadiPeyhani¹,

Hafner²,

Sveshnikova³

et al. 2021

Preprint

View full text Add to dashboard Cite

Metabolic "dark matter" describes currently unknown metabolic processes, which form a blind spot in our general understanding of metabolism and slow down the development of biosynthetic cell factories and naturally derived pharmaceuticals. Mapping the dark matter of metabolism remains an open challenge, one that can be addressed globally and systematically by existing computational solutions. In this work, we present ATLASx, a biochemical database of 1.5 million metabolic compounds and 490 generalized enzymatic reactions that can map both known and unknown metabolic processes. Using ATLASx, we were able to predict over 5 million reactions and integrate nearly 2 million naturally and synthetically-derived compounds into the global metabolic network. ATLASx is available to researchers online as a powerful tool that supports the analysis of novel biochemical pathways and evaluates the biochemical vicinity of molecule classes (https://lcsb-databases.epfl.ch/Atlas2).

show abstract

Section: Introductionmentioning

confidence: 99%

ATLASx: a computational map for the exploration of biochemical space

MohammadiPeyhani¹,

Hafner²,

Sveshnikova³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…To address the challenge of enzymatic synthesis pathway planning, a number of computational tools have been developed for general purpose bioretrosynthesis planning, [18][19][20][21][22] enzyme selection, 23,24 metabolic pathway exploration, 18,25 and reaction rule extraction 26,27 in recent years. The tools usually extract the catalyzed transformation from a known reaction by identifying the reactive center, coding the changes of atoms and bonds into a reaction rule, and scoring the feasibility of a new substrate undergoing the same transformation on a set of criteria.…”

Section: Introductionmentioning

confidence: 99%

“…Whereas some tools consider a reaction feasible if it suffices a reaction rule at a desired level of specificity, 21,25 others score the feasibility of a transformation based on chemical similarity to known reactions or substrates via fingerprint vectors. 20,[22][23][24] However, methods relying on similarity or reaction rule specificity lack the distinction between generalist and selective specialist enzymes, i.e. they miss a description of enzyme promiscuity, as pointed out by Jeffryes et al recently.…”

Section: Introductionmentioning

confidence: 99%

EHreact: Extended Hasse Diagrams for the Extraction and Scoring of Enzymatic Reaction Templates

Heid

Goldman²,

Sankaranarayanan³

et al. 2021

Preprint

View full text Add to dashboard Cite

Data-driven computer-aided synthesis planning utilizing organic or biocatalyzed reactions from large databases has gained increasing interest in the last decade, sparking the development of numerous tools to extract, apply and score general reaction templates. The generation of reaction rules for enzymatic reactions is especially challenging, since substrate promiscuity varies between enzymes, causing the optimal levels of rule specificity and optimal number of included atoms to differ between enzymes. This complicates an automated extraction from databases and has promoted the creation of manually curated reaction rule sets. Here we present EHreact, a purely data-driven open-source software tool to extract and score reaction rules from sets of reactions known to be catalyzed by an enzyme at appropriate levels of specificity without expert knowledge. EHreact extracts and groups reaction rules into tree-like structures, Hasse diagrams, based on common substructures in the imaginary transition structures. Each diagram can be utilized to output a single or a set of reaction rules, as well as calculate the probability of a new substrate to be processed by the given enzyme by inferring information about the reactive site of the enzyme from the known reactions and their grouping in the template tree. EHreact heuristically predicts the activity of a given enzyme on a new substrate, outperforming current approaches in accuracy and functionality.

show abstract

“…Briefly, from a given target compound and a given chassis strain, the first step consists of finding metabolic reactions that are heterologous to the chassis and link the target compound to the native metabolites of the host organism. This step is carried out by retrosynthesis software [8][9][10][11][12][13] and requires the use of reaction rules 14 if one wishes to search for novel pathways or find pathways that produce unnatural target compounds. The result of retrosynthesis software tools is a metabolic map and there is a need in a second step to enumerate the pathways linking the chassis metabolites to the target.…”

Section: Introductionmentioning

confidence: 99%

Galaxy-SynBioCAD: Synthetic Biology Design Automation tools in Galaxy workflows

Duigou

Hérisson

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Many computer-aided design tools are available for synthetic biology and metabolic engineering. Yet, these tools can be difficult to apprehend, sometimes requiring a level of expertise that limits their use by a wider community. Furthermore, some of the tools, although complementary, rely on different input and output formats and cannot communicate with one another. Scientific workflows address these shortcomings while offering a novel design strategy. Among the workflow systems available, Galaxy is a web-based platform for performing findable and accessible data analyses for all scientists regardless of their informatics expertise, along with interoperable and reproducible computations regardless of the particular platform that is being used.Here, we introduce the Galaxy-SynBioCAD a portal, the first Galaxy toolshed for synthetic biology and metabolic engineering. It allows one to easily create workflows or use those already developed by the community. The portal is a growing community effort where developers can add new tools and users can evaluate the tools performing design for their specific projects. The tools and workflows currently shared on the Galaxy-SynBioCAD portal cover an end-to-end metabolic pathway design process from the selection of strain and target to the calculation of DNA parts to be assembled to build libraries of strains to be engineered to produce the target.Standard formats are used throughout to enforce the compatibility of the tools. These include SBML for strain and pathway and SBOL for genetic layouts. The portal has been benchmarked on a galaxy-synbiocad.org 81 literature pathways, overall, we find we have a 65% (and 88%) success rate in retrieving the literature pathways among the top 10 (50) pathways predicted and generated by the workflows.

show abstract

Reinforcement Learning for Bioretrosynthesis

Cited by 107 publications

References 61 publications

ATLASx: a computational map for the exploration of biochemical space

ATLASx: a computational map for the exploration of biochemical space

EHreact: Extended Hasse Diagrams for the Extraction and Scoring of Enzymatic Reaction Templates

Galaxy-SynBioCAD: Synthetic Biology Design Automation tools in Galaxy workflows

Contact Info

Product

Resources

About