Deep learning for single-molecule science

Albrecht, Tim; Slabaugh, Greg; Alonso, Eduardo; Al-Arif, SM Masudur R

doi:10.1088/1361-6528/aa8334

Cited by 55 publications

(55 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Deep learning 16 is a machine-learning development that has been used to extract features and/or detect objects from different types of datasets for classification problems including base-calling in single-molecule analysis 17,18 . Convolutional neural network (CNN) layers are a powerful component of deep learning useful for learning patterns within complex data.…”

mentioning

confidence: 99%

Deep-Channel uses deep neural networks to detect single-molecule events from patch-clamp data

Celik¹,

O’Brien²,

Brennan³

et al. 2020

Commun Biol

View full text Add to dashboard Cite

Single-molecule research techniques such as patch-clamp electrophysiology deliver unique biological insight by capturing the movement of individual proteins in real time, unobscured by whole-cell ensemble averaging. The critical first step in analysis is event detection, so called "idealisation", where noisy raw data are turned into discrete records of protein movement. To date there have been practical limitations in patch-clamp data idealisation; high quality idealisation is typically laborious and becomes infeasible and subjective with complex biological data containing many distinct native single-ion channel proteins gating simultaneously. Here, we show a deep learning model based on convolutional neural networks and long short-term memory architecture can automatically idealise complex single molecule activity more accurately and faster than traditional methods. There are no parameters to set; baseline, channel amplitude or numbers of channels for example. We believe this approach could revolutionise the unsupervised automatic detection of single-molecule transition events in the future.

show abstract

mentioning

confidence: 99%

Deep-Channel uses deep neural networks to detect single-molecule events from patch-clamp data

Celik¹,

O’Brien²,

Brennan³

et al. 2020

Commun Biol

View full text Add to dashboard Cite

show abstract

“…Understanding the electronic quantum transport through organic materials is not straightforward. However, with the advent of machine learning and multivariate analysis techniques, vital information regarding the molecular conformation can be extracted from the stochastic peaks in the measured conductance 8 , where the deduced information is similar to that reported by optical spectroscopy techniques a) Electronic mail: aisakovic@colgate.edu. iregx137@gmail.com such as Fourier transform infrared (FTIR) and Raman.…”

Section: Introductionmentioning

confidence: 88%

Detection of some amino acids with modulation-doped and surface-nanoengineered GaAs Schottky P-I-N diodes

Alkhidir

Jaoudé

Gater

et al. 2020

Journal of Vacuum Science &Amp; Technology B, Nanotechnology and Microelectronics: Materials, Processing, Measurement, and Phen

View full text Add to dashboard Cite

Most current techniques for analyzing amino acids require substantial instrumentation and significant sample preprocessing. In this study, we designed, fabricated and tested a scalable diode-based microdevice that allows for direct sensing of amino acids. The device is based on modulation-doped GaAs heterostructure with a Schottky contact on one side. The relatively high mobility and relatively small dielectric constant of GaAs are naturally helpful in this problem. We also paid attention to a proper etching procedure allowing for substantial modification of the surface properties, thereby further boosting the sensing performance. Transport data (I-V, differential conductance) are presented for three qualitatively different classes of amino acids (i.e. non-polar with aliphatic R-group, polar uncharged R-group and charged R-group) with glycine, cysteine and histidine as specific examples, respectively. The conductance for the GaAsamino acid interface measured using scanning tunneling microscope (STM) was previously reported to have distinct spectral features. In this paper, we show that measuring the differential conductance of GaAs diode whose surface is in direct contact with an aqueous solution of amino acid, is a simple methodology to access useful information, previously available only through sophisticated and equipment-demanding STM and molecular electronics approaches. Density functional theory (DFT) calculations were used to examine which adsorption processes were likely responsible for the observed surface conductance modification. Lastly, in future and ongoing work, we illustrate how it might be possible to employ standard multivariate data analysis techniques to reliably identify distinct (95%) single amino acid specific features in near-ambient differential conductance data. Research questions we address here are: (1) is it possible to distinguish between different amino acids in contact with the same GaAs interface, based on I-V curves?; (2) how are partial charges specific to individual amino acids?; (3) what are the likely binding mechanisms (individual amino acid to GaAs)?; and (4) to what extent one can use differential conductance data? We show how it is possible to distinguish between separate aqueous solutions of the amino acids in in-This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

show abstract

“… 21 , 22 A recent study showed that CNNs perform well on simulated current traces from an STM tunnel junction. 23 For comparison, DNA bases can be accurately determined from current levels using recurrent neural networks. 24 However, our goal is fundamentally different, as we are trying to identify the pattern encoded on the DNA secondary structure from a variable nanopore system.…”

Section: Methodsmentioning

confidence: 99%

QuipuNet: Convolutional Neural Network for Single-Molecule Nanopore Sensing

2018

View full text Add to dashboard Cite

Nanopore sensing is a versatile technique for the analysis of molecules on the single-molecule level. However, extracting information from data with established algorithms usually requires time-consuming checks by an experienced researcher due to inherent variability of solid-state nanopores. Here, we develop a convolutional neural network (CNN) for the fully automated extraction of information from the time-series signals obtained by nanopore sensors. In our demonstration, we use a previously published data set on multiplexed single-molecule protein sensing. The neural network learns to classify translocation events with greater accuracy than previously possible, while also increasing the number of analyzable events by a factor of 5. Our results demonstrate that deep learning can achieve significant improvements in single molecule nanopore detection with potential applications in rapid diagnostics.

show abstract

Deep learning for single-molecule science

Cited by 55 publications

References 52 publications

Deep-Channel uses deep neural networks to detect single-molecule events from patch-clamp data

Deep-Channel uses deep neural networks to detect single-molecule events from patch-clamp data

Detection of some amino acids with modulation-doped and surface-nanoengineered GaAs Schottky P-I-N diodes

QuipuNet: Convolutional Neural Network for Single-Molecule Nanopore Sensing

Contact Info

Product

Resources

About