DeepNano: Deep recurrent neural networks for base calling in MinION nanopore reads

By Vladimír Boza, Broňa Brejová, Tomás Vinar. 2016

235 citations recorded by the source corpus.

The MinION device by Oxford Nanopore produces very long reads (reads over 100 kBp were reported); however it suffers from high sequencing error rate. We present an open-source DNA base caller based on deep recurrent neural networks and show that the accuracy of base calling is much dependent on the underlying software and can be improved by considering modern machine learning methods. By employing carefully crafted recurrent neural networks, our tool significantly improves base calling accuracy on data from R7.3 version of the platform compared to the default base caller supplied by the manufacturer. On R9 version, we achieve results comparable to Nanonet base caller provided by Oxford Nanopore. Availability of an open source tool with high base calling accuracy will be useful for development of new applications of the MinION device, including infectious disease detection and custom target enrichment during sequencing.

Subfields: Biomedical Engineering, Molecular Biology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biomedical Engineering

In Molecular Biology

Citations in the corpus

Opportunities and obstacles for deep learning in biology and medicine (2017, methodology): “However, because RNNs have been applied to base calls for the Oxford Nanopore long-read sequencer with some success [255] (discussed further in the next section), one day the entire pipeline, from denoising of through functional classification, may be combined into one step by using powerful LSTMs [”
Opportunities and challenges in long-read sequencing data analysis (2020, background): “Basecalling of nanopore reads is an area of active research, where algorithms are quickly evolving (neural networks have supplanted HMMs, and various neural networks structures are being tested [44]) as are the chemistries for which they are trained.”
Piercing the dark matter: bioinformatics of long-range sequencing and mapping (2018, background): “Substantial improvements have already been made for Oxford Nanopore base calling in the past few years, and today’s leading approaches can achieve >90% average accuracy using recurrent neural net approaches 49 and other statistical learning techniques 50 .”
From squiggle to basepair: computational approaches for improving nanopore sequencing read accuracy (2018, methodology): “3 data, Metrichor called 1D reads with an accuracy of 70 – 71% and Nanocall with an accuracy of 68%, whereas DeepNano reached accuracies of up to 77% [29, 35].”
Predicting enhancer-promoter interaction from genomic sequence with deep neural networks (2016, background): “Since then, these and other more elaborate deep learning architectures have been used in a variety of tasks in computational genomics, including base-calling [28], identification of transcription initiation sites [27], prediction of gene expression [25,26], and prediction of protein contact from ami”