Blog
About

0
views
0
recommends
+1 Recommend
0 collections
    0
    shares
      • Record: found
      • Abstract: found
      • Article: found
      Is Open Access

      MinCall - MinION end2end convolutional deep learning basecaller

      Preprint

      , ,

      Read this article at

      Bookmark
          There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.

          Abstract

          The Oxford Nanopore Technologies's MinION is the first portable DNA sequencing device. It is capable of producing long reads, over 100 kBp were reported. However, it has significantly higher error rate than other methods. In this study, we present MinCall, an end2end basecaller model for the MinION. The model is based on deep learning and uses convolutional neural networks (CNN) in its implementation. For extra performance, it uses cutting edge deep learning techniques and architectures, batch normalization and Connectionist Temporal Classification (CTC) loss. The best performing deep learning model achieves 91.4% median match rate on E. Coli dataset using R9 pore chemistry and 1D reads.

          Related collections

          Most cited references 6

          • Record: found
          • Abstract: not found
          • Article: not found

          Deep Residual Learning for Image Recognition

            Bookmark
            • Record: found
            • Abstract: found
            • Article: not found

            A complete bacterial genome assembled de novo using only nanopore sequencing data.

            We have assembled de novo the Escherichia coli K-12 MG1655 chromosome in a single 4.6-Mb contig using only nanopore data. Our method has three stages: (i) overlaps are detected between reads and then corrected by a multiple-alignment process; (ii) corrected reads are assembled using the Celera Assembler; and (iii) the assembly is polished using a probabilistic model of the signal-level data. The assembly reconstructs gene order and has 99.5% nucleotide identity.
              Bookmark
              • Record: found
              • Abstract: found
              • Article: not found

              A first look at the Oxford Nanopore MinION sequencer.

              Oxford Nanopore's third-generation single-molecule sequencing platform promises to decrease costs for reagents and instrumentation. After a 2-year hiatus following the initial announcement, the first devices have been released as part of an early access program. We explore the performance of this platform by resequencing the lambda phage genome, and amplicons from a snake venom gland transcriptome. Although the handheld MinION sequencer can generate more than 150 megabases of raw data in one run, at most a quarter of the resulting reads map to the reference, with less than average 10% identity. Much of the sequence consists of insertion/deletion errors, or is seemingly without similarity to the template. Using the lambda phage data as an example, although the reads are long, averaging 5 kb, at best 890 ± 1932 bases per mapped read could be matched to the reference without soft clipping. In the course of a 36 h run on the MinION, it was possible to resequence the 48 kb lambda phage reference at 16× coverage. Currently, substantially larger projects would not be feasible using the MinION. Without increases in accuracy, which would be required for applications such as genome scaffolding and phasing, the current utility of the MinION appears limited. Library preparation requires access to a molecular laboratory, and is of similar complexity and cost to that of other next-generation sequencing platforms. The MinION is an exciting step in a new direction for single-molecule sequencing, though it will require dramatic decreases in error rates before it lives up to its promise.
                Bookmark

                Author and article information

                Journal
                22 April 2019
                Article
                1904.10337

                http://arxiv.org/licenses/nonexclusive-distrib/1.0/

                Custom metadata
                2nd international workshop on deep learning for precision medicine, ECML-PKDD 2017
                q-bio.GN cs.LG

                Artificial intelligence, Genetics

                Comments

                Comment on this article