Predicting drug properties with parameter-free machine learning: pareto-optimal embedded modeling (POEM)

Brereton, Andrew E.; MacKinnon, Stephen S.; Safikhani, Zhaleh; Reeves, Shawn; Alwash, Sana; Shahani, Vijay; Windemuth, Andreas

doi:10.1088/2632-2153/ab891b

Cited by 9 publications

(4 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another example is Pareto-Optimal Embedded Modeling (POEM), a non-parametric, supervised machine learning algorithm developed to generate reliable predictive models without need for optimization. POEM’s predictive strength is obtained by combining multiple different representations of molecular structures [ 21 ].…”

Section: Introductionmentioning

confidence: 99%

Extended similarity indices: the benefits of comparing more than two objects simultaneously. Part 1: Theory and characteristics†

et al. 2021

View full text Add to dashboard Cite

Quantification of the similarity of objects is a key concept in many areas of computational science. This includes cheminformatics, where molecular similarity is usually quantified based on binary fingerprints. While there is a wide selection of available molecular representations and similarity metrics, there were no previous efforts to extend the computational framework of similarity calculations to the simultaneous comparison of more than two objects (molecules) at the same time. The present study bridges this gap, by introducing a straightforward computational framework for comparing multiple objects at the same time and providing extended formulas for as many similarity metrics as possible. In the binary case (i.e. when comparing two molecules pairwise) these are naturally reduced to their well-known formulas. We provide a detailed analysis on the effects of various parameters on the similarity values calculated by the extended formulas. The extended similarity indices are entirely general and do not depend on the fingerprints used. Two types of variance analysis (ANOVA) help to understand the main features of the indices: (i) ANOVA of mean similarity indices; (ii) ANOVA of sum of ranking differences (SRD). Practical aspects and applications of the extended similarity indices are detailed in the accompanying paper: Miranda-Quintana et al. J Cheminform. 2021. 10.1186/s13321-021-00504-4. Python code for calculating the extended similarity metrics is freely available at: https://github.com/ramirandaq/MultipleComparisons.

show abstract

Section: Introductionmentioning

confidence: 99%

Extended similarity indices: the benefits of comparing more than two objects simultaneously. Part 1: Theory and characteristics†

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Within the Ligand Express platform, the Pareto‐Optimal Embedded Modelling (POEM) algorithm was then used to predict the activity of TEDs in key physicochemical models. 30 A POEM classifier model for blood brain barrier penetration predicted that all TEDs would pass the barrier (0.87–0.99), they were likely not to be a substrate of P‐glycoprotein (0.13–0.40), and that all were likely to be absorbed within the intestine (0.92–0.99). A POEM model assessing likelihood of Ames mutagenicity (0.44–0.60) did not classify compounds as mutagenic, with predictions near a random value of 0.5.…”

Section: Resultsmentioning

confidence: 99%

“…Within the Ligand Express platform, the Pareto-Optimal Embedded Modelling (POEM) algorithm was then used to predict the activity of TEDs in key physicochemical models. 30 A POEM classifier model for blood brain barrier penetration predicted that all TEDs would pass the barrier (0.87-0.99), they were likely not to be a substrate of P-glycoprotein (0.13-0.40), and that all all compounds to be within an order of magnitude to PFI-3, suggesting that all would have comparable solubility. Thus, these physicochemical properties bode well for the potential development of these compounds as candidate therapeutic agents going forward.…”

Section: The Drug-like Properties Of Tedsmentioning

confidence: 99%

Next‐generation bromodomain inhibitors of the SWI/SNF complex enhance DNA damage and cell death in glioblastoma

Yang,

He,

Wang

et al. 2023

J Cellular Molecular Medi

Self Cite

View full text Add to dashboard Cite

Glioblastoma (GBM) is an aggressive brain cancer with a poor prognosis. While surgical resection is the primary treatment, adjuvant temozolomide (TMZ) chemotherapy and radiotherapy only provide slight improvement in disease course and outcome. Unfortunately, most treated patients experience recurrence of highly aggressive, therapy‐resistant tumours and eventually succumb to the disease. To increase chemosensitivity and overcome therapy resistance, we have modified the chemical structure of the PFI‐3 bromodomain inhibitor of the BRG1 and BRM catalytic subunits of the SWI/SNF chromatin remodelling complex. Our modifications resulted in compounds that sensitized GBM to the DNA alkylating agent TMZ and the radiomimetic bleomycin. We screened these chemical analogues using a cell death ELISA with GBM cell lines and a cellular thermal shift assay using epitope tagged BRG1 or BRM bromodomains expressed in GBM cells. An active analogue, IV‐129, was then identified and further modified, resulting in new generation of bromodomain inhibitors with distinct properties. IV‐255 and IV‐275 had higher bioactivity than IV‐129, with IV‐255 selectively binding to the bromodomain of BRG1 and not BRM, while IV‐275 bound well to both BRG1 and BRM bromodomains. In contrast, IV‐191 did not bind to either bromodomain or alter GBM chemosensitivity. Importantly, both IV‐255 and IV‐275 markedly increased the extent of DNA damage induced by TMZ and bleomycin as determined by nuclear γH2AX staining. Our results demonstrate that these next‐generation inhibitors selectively bind to the bromodomains of catalytic subunits of the SWI/SNF complex and sensitize GBM to the anticancer effects of TMZ and bleomycin. This approach holds promise for improving the treatment of GBM.

show abstract

“…The topic of ligand representations has been central to computational drug discovery tasks dating back to the earliest bioactivity models based on expert‐defined molecular descriptors (Craig, 1984). Outside the domain of DTI models, the relationship between ligand representations and performance of predictive tasks has been thoroughly explored, benchmarked, and reviewed (Banegas‐Luna, Cerón‐Carrasco, & Pérez‐Sánchez, 2018; Brereton et al., 2020; Jiang et al., 2021; Qing et al., 2014; Riniker & Landrum, 2013; Stepišnik, Škrlj, Wicker, & Kocev, 2021), generally noting that different bioactivity modeling and similarity problems benefit from different forms of chemical representation. Notably, the modeling toolkit DeepChem 2.5.0 offers over 40 different molecule featurizers that can be evaluated in different combinations with machine learning algorithms to identify optimal bioactivity models for specific tasks (Ramsundar et al., 2019).…”

Section: Deep Learning Models For Drug‐target Interaction Predictionsmentioning

confidence: 99%

Proteome‐Scale Drug‐Target Interaction Predictions: Approaches and Applications

MacKinnon¹,

Tonekaboni²,

Windemuth³

2021

Current Protocols

Self Cite

View full text Add to dashboard Cite

Drug‐Target interaction predictions are an important cornerstone of computer‐aided drug discovery. While predictive methods around individual targets have a long history, the application of proteome‐scale models is relatively recent. In this overview, we will provide the context required to understand advances in this emerging field within computational drug discovery, evaluate emerging technologies for suitability to given tasks, and provide guidelines for the design and implementation of new drug‐target interaction prediction models. We will discuss the validation approaches used, and propose a set of key criteria that should be applied to evaluate their validity. We note that we find widespread deficiencies in the existing literature, making it difficult to judge the practical effectiveness of some of the techniques proposed from their publications alone. We hope that this review may help remedy this situation and increase awareness of several sources of bias that may enter into commonly used cross‐validation methods. © 2021 Cyclica Inc. Current Protocols published by Wiley Periodicals LLC.

show abstract

Predicting drug properties with parameter-free machine learning: pareto-optimal embedded modeling (POEM)

Cited by 9 publications

References 32 publications

Extended similarity indices: the benefits of comparing more than two objects simultaneously. Part 1: Theory and characteristics†

Extended similarity indices: the benefits of comparing more than two objects simultaneously. Part 1: Theory and characteristics†

Next‐generation bromodomain inhibitors of the SWI/SNF complex enhance DNA damage and cell death in glioblastoma

Proteome‐Scale Drug‐Target Interaction Predictions: Approaches and Applications

Contact Info

Product

Resources

About