Home

Awesome

ONT_logo

We have a new bioinformatic resource that largely replaces the functionality of this project! See our new repository here: https://github.com/nanoporetech/bonito

This repository is now unsupported and we do not recommend its use. Please contact Oxford Nanopore: support@nanoporetech.com for help with your application if it is not possible to upgrade to our new resources, or we are missing key features.


Flappie

Overview

Basecall Fast5 reads using flip-flop basecalling.

For run-length encoded basecalling, see Runnie

Features

Getting Started

Input and Output

Installation

Flappie has been tested on Ubuntu 16.04.5 LTS. Other systems may be compatible.

Flappie models and other large resources are stored using git lfs and this extension must be installed to successfully clone the repository.

git clone https://github.com/nanoporetech/flappie
cd flappie
make flappie

An alternative location for the HDF5 library, for example one installed by brew, can be specified as:

hdf5Root=/usr/local/ make flappie

Compilation From Source

Flappie has the following dependences

On Debian based systems, the following packages are sufficient (tested Ubuntu 14.04 and 16.04)

Usage

#  ! It is highly recommended that OpenBLAS is run in single threaded mode
export OPENBLAS_NUM_THREADS=1
#  List available models
flappie --model help
#  Basecall reads directory
flappie reads/ > basecalls.fq
#  Basecall using a different model
flappie --model r941_5mC reads/ > basecalls.fq
#  Output to SAM (not compatible with modification calls)
flappie --format sam reads/ > basecalls.sam
#  Output to BAM (not compatible with modification calls)
flappie --format sam reads | samtools view -Sb - > basecalls.bam
#  Dump trace data
flappie --trace trace.hdf5 reads > basecalls.fq
#  Basecall in parallel
find reads -name \*.fast5 | parallel -P $(nproc) -X flappie > basecalls.fq
#  Dump trace in parallel.  One trace per parallel process.
find reads -name \*.fast5 | parallel -P $(nproc) -X flappie --trace trace_{%}.hdf5 {} > basecalls.fq

Detecting RNA

Calling RNA differs from DNA in two respects:

  1. Output sequences must be reversed, since the molecule is sequenced in reverse (3' to 5')
  2. 'Delta scaling' must be used.

Direct RNA flip-flop is trained in a different manner from previous models, meaning that it is invariant to shifts in current and does not normalise the range of each read. The difference in normalisation from standard flip-flop improves accuracy for short reads, or those with unusual sequence composition.

flappie --model r941_rna002 --reverse --delta 1.0 reads/ > basecalls.fq

Trace viewer

A basic trace viewer is supplied with Flappie, supporting trace output for both Flappie and Guppy.

nanopore trace with 5 bases

#  Set up
virtualenv -p python3 venv
source venv/bin/activate
pip install --upgrade pip
pip install -r misc/trace_requirements.txt

#  View a trace -- Flappie trace output
misc/trace_flipflop.py trace.hdf5

#  View a trace -- Guppy trace output
misc/trace_flipflop.py guppy_trace.fast5

#  View a trace -- Guppy trace output, additional analysis
misc/trace_flipflop.py --analysis 1 guppy_trace.fast5

Help

Licence and Copyright

(c) 2018 Oxford Nanopore Technologies Ltd.

Flappie is distributed under the terms of the Oxford Nanopore Technologies, Ltd. Public License, v. 1.0. If a copy of the License was not distributed with this file, You can obtain one at http://nanoporetech.com

The vectorised math functions used by Flappie src/sse_mathfun.h are from http://gruntthepeon.free.fr/ssemath/ and the original version of this file is under the 'zlib' licence. See the top of src/sse_mathfun.h for details.

FAQs

Compilation failures

Git LFS missing

If you encounter compilation failures of the following form, the repository was cloned without git lfs and the model files are missing.

/home/ubuntu/mounted/extensionBonusFlappie/flappie/src/models/flipflop_r941native.h:1:1: error: unknown type name ‘version’
version https://git-lfs.github.com/spec/v1
^~~~~~~
/home/ubuntu/mounted/extensionBonusFlappie/flappie/src/models/flipflop_r941native.h:1:14: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘:’ token
version https://git-lfs.github.com/spec/v1
              ^
/home/ubuntu/mounted/extensionBonusFlappie/flappie/src/models/flipflop_r941native.h:2:12: error: invalid suffix "fa49fa0ea6c67806f69fd6ba42a7dd935390b86f98a615ee03e167d22583" on integer constant
oid sha256:3979fa49fa0ea6c67806f69fd6ba42a7dd935390b86f98a615ee03e167d22583
            ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
In file included from /home/ubuntu/mounted/extensionBonusFlappie/flappie/src/networks.c:4:0:
/home/ubuntu/mounted/extensionBonusFlappie/flappie/src/models/flipflop_r10Cpcr.h:2:12: error: invalid suffix "b2fe5fd1c3d9646e7a6ea76e646beb50ad6a5fe17e5da0c76c13bd907cb4" on floating constant
oid sha256:83e2b2fe5fd1c3d9646e7a6ea76e646beb50ad6a5fe17e5da0c76c13bd907cb4
            ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

RedHat and Centos systems

Redhat and Centos systems install openblas as a separate library, leading to errors of the following form since blas cannot be found:

[ 84%] Linking C executable flappie
/usr/bin/ld: cannot find -lblas
collect2: ld returned 1 exit
status make[4]: *** [flappie] Error 1

While neither RedHat nor Centos are tested or supported, we have added an argument to assist compiling flappie on these platforms.

make openblasRedHat=1 flappie

High system load

Extremely high system load can arise when Flappie is run using parallel and OpenBLAS is used in multi-threaded mode. Running in this manner is harmful to overall throughput and it is recommended that OpenBLAS is used in single-threaded manner, which can be enable by setting the OPENBLAS_NUM_THREADS environmental variable to 1 (see top of Usage for an example of how to do this).

Installing git-lfs

From git lfs, the installation instructions for Debian based systems, like Ubuntu, are:

curl -s https://packagecloud.io/install/repositories/github/git-lfs/script.deb.sh | sudo bash
sudo apt-get install git-lfs
git lfs install

Methylation and other modifications

Flappie currently only calls 5mC methylation in CpG contexts. Calling other modifications, or 5mC in other contexts, is not currently supported.

Methylated calls are currently represented as a 'Z' base in the output -- this is likely to break down-stream tools and may not be final format in which this modification information is represented. Outputting methylation from Flappie is enabled for early adopters to think about how this information may be used; please do not rely on this particular representation of modifications as it may change in the future. Particularly, the SAM output when modification calling is enabled does not conform to the published specification for that for format.

See https://github.com/nanoporetech/flappie/issues/11 and https://github.com/samtools/hts-specs/issues/362 for more details of the issues involved.

Platform support

The models contained contained in Flappie are trained using data from the MinION platform. Use on other platforms is not supported, although they may generalise to reads from the GridION platform due to the similarity of the hardware.

Quality scores

The quality currently produced for FASTQ and SAM output are derived directly from the probabilistic model output by the Flappie model and have not been calibrated.

Trace file

The trace information is output as a block x state matrix, where the states are the flip (uppercase) and flop (lowercase) bases in the order ACGTacgt or ACGTZacgtz for methylated calls. The probabilities for each state are normalised into the range 0..255 and then represented by an unsigned 8bit integer. Due to rounding, the sum of encoded probabilities for each block may not equal 255.

Abbreviations

References and Supporting Information

Research Release

Research releases are provided as technology demonstrators to provide early access to features or stimulate Community development of tools. Support for this software will be minimal and is only provided directly by the developers. Feature requests, improvements, and discussions are welcome and can be implemented by forking and pull requests. However much as we would like to rectify every issue and piece of feedback users may have, the developers may have limited resource for support of this software. Research releases may be unstable and subject to rapid iteration by Oxford Nanopore Technologies.