Evolving the Olfactory System with Machine Learning

The convergent evolution of the fly and mouse olfactory system led us to ask whether the anatomic connectivity and functional logic in vivo would evolve in artificial neural networks constructed to perform olfactory tasks. Artificial networks trained to classify odor identity recapitulate the connectivity inherent in the olfactory system. Input units are driven by a single receptor type, and units driven by the same receptor converge to form a glomerulus. Glomeruli exhibit sparse, unstructured connectivity to a larger, expansion layer. When trained to both classify odor and impart innate valence on odors, the network develops independent pathways for innate output and odor classification. Thus, artificial networks evolve even without the biological mechanisms necessary to build these systems in vivo, providing a rationale for the convergent evolution of olfactory circuits.

Download Full-text

Predicting Absorption-Distribution Properties of Neuroprotective Phosphine-Borane Compounds Using In Silico Modeling and Machine Learning

Molecules ◽

10.3390/molecules26092505 ◽

2021 ◽

Vol 26 (9) ◽

pp. 2505

Author(s):

Raheem Remtulla ◽

Sanjoy Kumar Das ◽

Leonard A. Levin

Keyword(s):

Machine Learning ◽

Neural Networks ◽

In Silico ◽

Disulfide Bonds ◽

Oral Absorption ◽

In Vivo Studies ◽

Drug Candidates ◽

In Silico Methods

Phosphine-borane complexes are novel chemical entities with preclinical efficacy in neuronal and ophthalmic disease models. In vitro and in vivo studies showed that the metabolites of these compounds are capable of cleaving disulfide bonds implicated in the downstream effects of axonal injury. A difficulty in using standard in silico methods for studying these drugs is that most computational tools are not designed for borane-containing compounds. Using in silico and machine learning methodologies, the absorption-distribution properties of these unique compounds were assessed. Features examined with in silico methods included cellular permeability, octanol-water partition coefficient, blood-brain barrier permeability, oral absorption and serum protein binding. The resultant neural networks demonstrated an appropriate level of accuracy and were comparable to existing in silico methodologies. Specifically, they were able to reliably predict pharmacokinetic features of known boron-containing compounds. These methods predicted that phosphine-borane compounds and their metabolites meet the necessary pharmacokinetic features for orally active drug candidates. This study showed that the combination of standard in silico predictive and machine learning models with neural networks is effective in predicting pharmacokinetic features of novel boron-containing compounds as neuroprotective drugs.

Download Full-text

In Silico Prediction of siRNA Ionizable-Lipid Nanoparticles in vivo Efficacy: Machine Learning Modeling Based on Formulation and Molecular Descriptors

10.20944/preprints202108.0254.v1 ◽

2021 ◽

Author(s):

Abdelkader A Metwally ◽

Amira A Nayel ◽

Rania M Hathout

Keyword(s):

Machine Learning ◽

Neural Networks ◽

Artificial Neural Networks ◽

In Silico ◽

Molecular Descriptors ◽

Lipid Nanoparticles ◽

Data Set ◽

In Vivo Efficacy ◽

Artificial Neural

In silico prediction of the in vivo efficacy of siRNA ionizable-lipid nanoparticles is desirable yet never achieved before. This study aims to computationally predict siRNA nanoparticles in vivo efficacy, which saves time and resources. A data set containing 120 entries was prepared by combining molecular descriptors of the ionizable lipids together with two nanoparticles formulation characteristics. Input descriptor combinations were selected by an evolutionary algorithm. Artificial neural networks, support vector machines and partial least squares regression were used for QSAR modeling. Depending on how the data set is split, two training sets and two external validation sets were prepared. Training and validation sets contained 90 and 30 entries respectively. The results showed the successful predictions of validation set log(dose) with R2val = 0.86 – 0.89 and 0.75 – 80 for validation sets one and two respectively. Artificial neural networks resulted in the best R2val for both validation sets. For predictions that have high bias, improvement of R2val from 0.47 to 0.96 was achieved by selecting the training set lipids lying within the applicability domain. In conclusion, in vivo performance of siRNA nanoparticles was successfully predicted by combining cheminformatics with machine learning techniques.

Download Full-text

Non-canonical odor coding ensures unbreakable mosquito attraction to humans

10.1101/2020.11.07.368720 ◽

2020 ◽

Author(s):

Meg A. Younger ◽

Margaret Herre ◽

Alison R. Ehrlich ◽

Zhongyan Gong ◽

Zachary N. Gilbert ◽

...

Keyword(s):

Olfactory System ◽

Receptor Gene ◽

Gene Families ◽

Receptor Expression ◽

Olfactory Neurons ◽

Odor Cues ◽

Chemosensory Receptor ◽

Single Receptor ◽

The Brain

SUMMARYFemale Aedes aegypti mosquitoes show strong innate attraction to humans. This chemosensory behavior is critical to species survival because females require a blood-meal to reproduce. Humans, the preferred host of Ae. aegypti, produce a complex blend of odor cues along with carbon dioxide (CO2) that attracts females ready to bite. Mosquitoes detect these cues with heteromeric ligand-gated ion channels encoded by three different chemosensory receptor gene families. A common theme in other species is that olfactory neurons express a single receptor that defines their chemical specificity and that they extend axons that converge upon dedicated glomeruli in the first sensory processing center in the brain. Such an organization permits the brain to segregate olfactory information and monitor activity of individual glomeruli to interpret what smell has been encountered. We have discovered that Ae. aegypti uses an entirely different organizational principle for its olfactory system. Using genetic strains that label subpopulations of olfactory neurons, we found that many neurons co-express multiple members of at least two of the chemosensory receptor families. This unexpected co-expression is functional, as assessed by in vivo calcium imaging showing that a given glomerulus is activated by multiple ligands detected by different receptor families. This has direct functional consequences for mosquito behavior. Mutant mosquitoes that cannot sense CO2 can be behaviorally activated by a volatile amine that stimulates the CO2 glomerulus. This non-canonical olfactory system organization featuring overlapping receptor expression may explain the female mosquito’s robust and “unbreakable’ attraction to humans.

Download Full-text

Attack and defence in cellular decision-making: lessons from machine learning

10.1101/366724 ◽

2018 ◽

Author(s):

Thomas J. Rademaker ◽

Emmanuel Bengio ◽

Paul François

Keyword(s):

Machine Learning ◽

Neural Networks ◽

Decision Making ◽

Critical Point ◽

Nonlinear Models ◽

Machine Learning Algorithms ◽

Immune Recognition ◽

Cellular Decision Making ◽

Decision Making Models

Machine learning algorithms can be fooled by small well-designed adversarial perturbations. This is reminiscent of cellular decision-making where ligands (called antagonists) prevent correct signalling, like in early immune recognition. We draw a formal analogy between neural networks used in machine learning and models of cellular decision-making (adaptive proofreading). We apply attacks from machine learning to simple decision-making models, and show explicitly the correspondence to antagonism by weakly bound ligands. Such antagonism is absent in more nonlinear models, which inspired us to implement a biomimetic defence in neural networks filtering out adversarial perturbations. We then apply a gradient-descent approach from machine learning to different cellular decision-making models, and we reveal the existence of two regimes characterized by the presence or absence of a critical point for the gradient. This critical point causes the strongest antagonists to lie close to the decision boundary. This is validated in the loss landscapes of robust neural networks and cellular decision-making models, and observed experimentally for immune cells. For both regimes, we explain how associated defence mechanisms shape the geometry of the loss landscape, and why different adversarial attacks are effective in different regimes. Our work connects evolved cellular decision-making to machine learning, and motivates the design of a general theory of adversarial perturbations, both for in vivo and in silico systems.

Download Full-text

MACHINE LEARNING FOR SOFTWARE

International Conference on Technics Technologies and Education ◽

10.15547/ictte.2019.04.061 ◽

2019 ◽

pp. 251-255

Author(s):

Shafagat Mahmudova

Keyword(s):

Machine Learning ◽

Neural Networks ◽

Soft Computing ◽

Computing Technology

The study machine learning for software based on Soft Computing technology. It analyzes Soft Computing components. Their use in software, their advantages and challenges are studied. Machine learning and its features are highlighted. The functions and features of neural networks are clarified, and recommendations were given.

Download Full-text

Nanosecond Photodynamics Simulations of a Cis-Trans Isomerization Are Enabled by Machine Learning

10.26434/chemrxiv.13047863 ◽

2020 ◽

Author(s):

Jingbai Li ◽

Patrick Reiser ◽

André Eberhard ◽

Pascal Friederich ◽

Steven Lopez

Keyword(s):

Machine Learning ◽

Neural Networks ◽

Excited State ◽

Adaptive Sampling ◽

Computational Cost ◽

Ground Truth ◽

Absolute Error ◽

Photochemical Reactions ◽

Computational Techniques ◽

Full Potential

Photochemical reactions are being increasingly used to construct complex molecular architectures with mild and straightforward reaction conditions. Computational techniques are increasingly important to understand the reactivities and chemoselectivities of photochemical isomerization reactions because they offer molecular bonding information along the excited-state(s) of photodynamics. These photodynamics simulations are resource-intensive and are typically limited to 1–10 picoseconds and 1,000 trajectories due to high computational cost. Most organic photochemical reactions have excited-state lifetimes exceeding 1 picosecond, which places them outside possible computational studies. Westermeyr et al. demonstrated that a machine learning approach could significantly lengthen photodynamics simulation times for a model system, methylenimmonium cation (CH2NH2+).We have developed a Python-based code, Python Rapid Artificial Intelligence Ab Initio Molecular Dynamics (PyRAI2MD), to accomplish the unprecedented 10 ns cis-trans photodynamics of trans-hexafluoro-2-butene (CF3–CH=CH–CF3) in 3.5 days. The same simulation would take approximately 58 years with ground-truth multiconfigurational dynamics. We proposed an innovative scheme combining Wigner sampling, geometrical interpolations, and short-time quantum chemical trajectories to effectively sample the initial data, facilitating the adaptive sampling to generate an informative and data-efficient training set with 6,232 data points. Our neural networks achieved chemical accuracy (mean absolute error of 0.032 eV). Our 4,814 trajectories reproduced the S1 half-life (60.5 fs), the photochemical product ratio (trans: cis = 2.3: 1), and autonomously discovered a pathway towards a carbene. The neural networks have also shown the capability of generalizing the full potential energy surface with chemically incomplete data (trans → cis but not cis → trans pathways) that may offer future automated photochemical reaction discoveries.

Download Full-text

Development of Synchrotron Footprinting at NSLS and NSLS-II

Protein and Peptide Letters ◽

10.2174/0929866526666181128125125 ◽

2019 ◽

Vol 26 (1) ◽

pp. 55-60 ◽

Cited By ~ 1

Author(s):

Jen Bohon

Keyword(s):

Solvent Accessibility ◽

Biological Systems ◽

In Vivo Studies ◽

Aqueous Samples ◽

Biological Mechanisms ◽

Macromolecular Assembly ◽

Structure And Dynamics ◽

Synchrotron Light ◽

Ideal Technique

Background: First developed in the 1990’s at the National Synchrotron Light Source, xray synchrotron footprinting is an ideal technique for the analysis of solution-state structure and dynamics of macromolecules. Hydroxyl radicals generated in aqueous samples by intense x-ray beams serve as fine probes of solvent accessibility, rapidly and irreversibly reacting with solvent exposed residues to provide a “snapshot” of the sample state at the time of exposure. Over the last few decades, improvements in instrumentation to expand the technology have continuously pushed the boundaries of biological systems that can be studied using the technique. Conclusion: Dedicated synchrotron beamlines provide important resources for examining fundamental biological mechanisms of folding, ligand binding, catalysis, transcription, translation, and macromolecular assembly. The legacy of synchrotron footprinting at NSLS has led to significant improvement in our understanding of many biological systems, from identifying key structural components in enzymes and transporters to in vivo studies of ribosome assembly. This work continues at the XFP (17-BM) beamline at NSLS-II and facilities at ALS, which are currently accepting proposals for use.

Download Full-text

Discretization and machine learning approximation of BSDEs with a constraint on the Gains-process

Monte Carlo Methods and Applications ◽

10.1515/mcma-2020-2080 ◽

2021 ◽

Vol 0 (0) ◽

Author(s):

Idris Kharroubi ◽

Thomas Lim ◽

Xavier Warin

Keyword(s):

Neural Network ◽

Machine Learning ◽

Neural Networks ◽

Differential Equations ◽

Numerical Experiments ◽

Optimization Problem ◽

Learning Approach ◽

The Neural Network ◽

Machine Learning Approach ◽

Mesh Grid

AbstractWe study the approximation of backward stochastic differential equations (BSDEs for short) with a constraint on the gains process. We first discretize the constraint by applying a so-called facelift operator at times of a grid. We show that this discretely constrained BSDE converges to the continuously constrained one as the mesh grid converges to zero. We then focus on the approximation of the discretely constrained BSDE. For that we adopt a machine learning approach. We show that the facelift can be approximated by an optimization problem over a class of neural networks under constraints on the neural network and its derivative. We then derive an algorithm converging to the discretely constrained BSDE as the number of neurons goes to infinity. We end by numerical experiments.

Download Full-text

A Machine Learning Approach as a Surrogate for a Finite Element Analysis: Status of Research and Application to One Dimensional Systems

Sensors ◽

10.3390/s21051654 ◽

2021 ◽

Vol 21 (5) ◽

pp. 1654

Author(s):

Poojitha Vurtur Badarinath ◽

Maria Chierichetti ◽

Fatemeh Davoudi Kakhki

Keyword(s):

Machine Learning ◽

Neural Networks ◽

Finite Element Analysis ◽

Finite Element ◽

Artificial Neural Networks ◽

Stress Distribution ◽

Maintenance Scheduling ◽

Element Analysis ◽

One Dimensional ◽

Artificial Neural

Current maintenance intervals of mechanical systems are scheduled a priori based on the life of the system, resulting in expensive maintenance scheduling, and often undermining the safety of passengers. Going forward, the actual usage of a vehicle will be used to predict stresses in its structure, and therefore, to define a specific maintenance scheduling. Machine learning (ML) algorithms can be used to map a reduced set of data coming from real-time measurements of a structure into a detailed/high-fidelity finite element analysis (FEA) model of the same system. As a result, the FEA-based ML approach will directly estimate the stress distribution over the entire system during operations, thus improving the ability to define ad-hoc, safe, and efficient maintenance procedures. The paper initially presents a review of the current state-of-the-art of ML methods applied to finite elements. A surrogate finite element approach based on ML algorithms is also proposed to estimate the time-varying response of a one-dimensional beam. Several ML regression models, such as decision trees and artificial neural networks, have been developed, and their performance is compared for direct estimation of the stress distribution over a beam structure. The surrogate finite element models based on ML algorithms are able to estimate the response of the beam accurately, with artificial neural networks providing more accurate results.

Download Full-text

Trigonometric Inference Providing Learning in Deep Neural Networks

Applied Sciences ◽

10.3390/app11156704 ◽

2021 ◽

Vol 11 (15) ◽

pp. 6704

Author(s):

Jingyong Cai ◽

Masashi Takemoto ◽

Yuming Qiu ◽

Hironori Nakajo

Keyword(s):

Machine Learning ◽

Neural Networks ◽

Deep Neural Networks ◽

Activation Function ◽

Trigonometric Approximation ◽

Model Parameters ◽

Training Algorithms ◽

Activation Functions ◽

Classical Training ◽

Sum Formula

Despite being heavily used in the training of deep neural networks (DNNs), multipliers are resource-intensive and insufficient in many different scenarios. Previous discoveries have revealed the superiority when activation functions, such as the sigmoid, are calculated by shift-and-add operations, although they fail to remove multiplications in training altogether. In this paper, we propose an innovative approach that can convert all multiplications in the forward and backward inferences of DNNs into shift-and-add operations. Because the model parameters and backpropagated errors of a large DNN model are typically clustered around zero, these values can be approximated by their sine values. Multiplications between the weights and error signals are transferred to multiplications of their sine values, which are replaceable with simpler operations with the help of the product to sum formula. In addition, a rectified sine activation function is utilized for further converting layer inputs into sine values. In this way, the original multiplication-intensive operations can be computed through simple add-and-shift operations. This trigonometric approximation method provides an efficient training and inference alternative for devices with insufficient hardware multipliers. Experimental results demonstrate that this method is able to obtain a performance close to that of classical training algorithms. The approach we propose sheds new light on future hardware customization research for machine learning.

Download Full-text