DeepOlf: Deep Neural Network Based Architecture for Predicting Odorants and Their Interacting Olfactory Receptors

Sharma, Anju; Rajnish, Kumar; Semwal, Rahul; Aier, Imlimaong; Tyagi, Pankaj; Varadwaj, Pritish Kumar

doi:10.1109/tcbb.2020.3002154

Cited by 25 publications

(18 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…1C). Sharma and others recently published a deep neural network model that predicts odor classification with comparable accuracy (AUROC 0.98) using 1622 chemical features and molecules drawn from internet databases ( 7 ); however, our results indicate that this degree of model complexity is not necessary to generate reliable predictions.…”

mentioning

confidence: 62%

Drawing the Borders of Olfactory Space

Mayhew

Arayata

Gerkin

et al. 2020

Preprint

View full text Add to dashboard Cite

In studies of vision and audition, stimuli can be chosen to span the visible or audible spectrum; in olfaction, the axes and boundaries defining the analogous odorous space are unknown. Features governing the physical transport of molecules to olfactory receptors are sufficient to reliably classify novel molecules as odorous or odorless (AUROC = 0.97). Applying this model to a database of all possible small organic molecules, we estimate that over 30 billion possible compounds are odorous, 6 orders of magnitude larger than current estimates of 10,000. Remarkably, nearly all transport-capable molecules are odorous, suggesting broad collective tuning of olfactory receptors. Defining the boundaries of odor perception will enable design of experiments that representatively sample olfactory space and efficient search for novel odor compounds.

show abstract

mentioning

confidence: 62%

Drawing the Borders of Olfactory Space

Mayhew

Arayata

Gerkin

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…To achieve this, we utilized an independent, held-out validation dataset from our curated interactions. We tested this validation dataset as an input query on two other prediction models, that is, ODORactor ( 65 ) and DeepOlf ( 66 ). Of note, the comparative analysis was only feasible for computing the model precision because of the functional limitations associated with DeepOlf ( 66 ) and ODORactor ( 65 ) ( Fig.…”

Section: Resultsmentioning

confidence: 99%

“…Importantly, all the aforementioned prediction engines also support modules of explainable artificial intelligence, allowing the user to decode the underlying features for the model classification/prediction at the atomic (odorant or nonodorant) and amino acid levels (ORs). Of note, machine/deep learning–based approaches have been used in the past to build prediction models for OR–odorant interactions as well as for the odorant/nonodorant classification ( 65 , 66 ). Notably, both ODORactor ( 65 ) as well as recently introduced DeepOlf ( 66 ) utilizes manually designed molecular descriptors for model building as well as for predictions, whereas, OdoriFy models utilize One-Hot encoding/BiLSTM-based techniques.…”

Section: Discussionmentioning

confidence: 99%

“…We tested this validation dataset as an input query on two other prediction models i.e. ODORactor (72) and DeepOlf (73). Of note, the comparative analysis was only feasible for computing the model precision due to the functional limitations associated with DeepOlf (73) and ODORactor (72) (Supplementary Figure 7).…”

Section: Odorify Synergizes With the Structure-based Or-ligand Interaction Methods And Outperforms Other Prediction Toolsmentioning

confidence: 99%

See 1 more Smart Citation

OdoriFy: A conglomerate of artificial intelligence–driven prediction engines for olfactory decoding

Gupta

Mittal

Agrawal

et al. 2021

Journal of Biological Chemistry

View full text Add to dashboard Cite

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

show abstract

“…Artificial intelligence (AI) is the computational design, development, and application of computer programs and algorithms that perform cognitive functions based on human intelligence traits, for example, anticipating, problem-solving, and learning ( Saxena et al, 2019 ; Sharma A. et al, 2020 ). AI techniques have the potential to accelerate the virtual screening, lead discovery and validation, etc .…”

Section: Identification Of Therapeutic Biomolecules Of Plants Through the Multi-omics Approachmentioning

confidence: 99%

Multi-Omics Approach in the Identification of Potential Therapeutic Biomolecule for COVID-19

Singh

Kumar

et al. 2021

Front. Pharmacol.

Self Cite

View full text Add to dashboard Cite

COVID-19 is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). It has a disastrous effect on mankind due to the contagious and rapid nature of its spread. Although vaccines for SARS-CoV-2 have been successfully developed, the proven, effective, and specific therapeutic molecules are yet to be identified for the treatment. The repurposing of existing drugs and recognition of new medicines are continuously in progress. Efforts are being made to single out plant-based novel therapeutic compounds. As a result, some of these biomolecules are in their testing phase. During these efforts, the whole-genome sequencing of SARS-CoV-2 has given the direction to explore the omics systems and approaches to overcome this unprecedented health challenge globally. Genome, proteome, and metagenome sequence analyses have helped identify virus nature, thereby assisting in understanding the molecular mechanism, structural understanding, and disease propagation. The multi-omics approaches offer various tools and strategies for identifying potential therapeutic biomolecules for COVID-19 and exploring the plants producing biomolecules that can be used as biopharmaceutical products. This review explores the available multi-omics approaches and their scope to investigate the therapeutic promises of plant-based biomolecules in treating SARS-CoV-2 infection.

show abstract

DeepOlf: Deep Neural Network Based Architecture for Predicting Odorants and Their Interacting Olfactory Receptors

Cited by 25 publications

References 34 publications

Drawing the Borders of Olfactory Space

Drawing the Borders of Olfactory Space

OdoriFy: A conglomerate of artificial intelligence–driven prediction engines for olfactory decoding

Multi-Omics Approach in the Identification of Potential Therapeutic Biomolecule for COVID-19

Contact Info

Product

Resources

About