jayhesselberth / f5c

Ultra-fast methylation calling and event alignment tool for nanopore sequencing data (supports CUDA acceleration)

f5c

An optimised re-implementation of the index, call-methylation and eventalign modules in Nanopolish. Given a set of basecalled Nanopore reads and the raw signals, f5c call-methylation detects the methylated cytosine and f5c eventalign aligns raw nanopore signals (events) to the reference k-mers. f5c can optionally utilise NVIDIA graphics cards for acceleration. f5c v1.2 onwards support the latest nanopore R10.4.1 chemistry.

First, the reads have to be indexed using f5c index. Then, invoke f5c call-methylation to detect methylated cytosine bases. Finally, you may use f5c meth-freq to obtain methylation frequencies. Alternatively, invoke f5c eventalign to perform event alignment. The results are almost the same as from nanopolish except a few differences due to floating point approximations.

Full Documentation : https://hasindu2008.github.io/f5c/docs/overview
Latest release : https://github.com/hasindu2008/f5c/releases/latest
Pre-print : https://doi.org/10.1101/756122
Publication : https://doi.org/10.1186/s12859-020-03697-x
Supplementary: nanopore_signal_alignment_supplementary_material.pdf
Talk Video : https://www.nvidia.com/en-us/on-demand/session/gtcspring21-s31391

Please cite the following when using f5c in your publications:

Gamaarachchi, H., Lam, C.W., Jayatilaka, G. et al. GPU accelerated adaptive banded event alignment for rapid comparative nanopore signal analysis. BMC Bioinformatics 21, 343 (2020). https://doi.org/10.1186/s12859-020-03697-x

@article{gamaarachchi2020gpu,
  title={GPU accelerated adaptive banded event alignment for rapid comparative nanopore signal analysis},
  author={Gamaarachchi, Hasindu and Lam, Chun Wai and Jayatilaka, Gihan and Samarakoon, Hiruna and Simpson, Jared T and Smith, Martin A and Parameswaran, Sri},
  journal={BMC bioinformatics},
  volume={21},
  number={1},
  pages={1--13},
  year={2020},
  publisher={BioMed Central}
}

Quick start

If you are a Linux user and want to quickly try out download the compiled binaries from the latest release. For example:

VERSION=v1.3
wget "https://github.com/hasindu2008/f5c/releases/download/$VERSION/f5c-$VERSION-binaries.tar.gz" && tar xvf f5c-$VERSION-binaries.tar.gz && cd f5c-$VERSION/
./f5c_x86_64_linux        # CPU version
./f5c_x86_64_linux_cuda   # cuda supported version

Binaries should work on most Linux distributions as the only dependency is zlib which is available by default on most distroibutions. For compiled binaries to work, your processor must support SSSE3 instructions or higher (processors after 2007 have these) and your operating system must have GLIBC 2.17 or higher (Linux distributions from 2014 onwards typically have this).

You can also use conda to install f5c as conda install f5c -c bioconda -c conda-forge.

Building from source

Building a release

Users are recommended to build from the latest release tar ball. You need a compiler that supports C++11. Quick example for Ubuntu :

sudo apt-get install libhdf5-dev zlib1g-dev   #install HDF5 and zlib development libraries
VERSION=v1.3
wget "https://github.com/hasindu2008/f5c/releases/download/$VERSION/f5c-$VERSION-release.tar.gz" && tar xvf f5c-$VERSION-release.tar.gz && cd f5c-$VERSION/
scripts/install-hts.sh  # download and compile the htslib
./configure
make                    # make cuda=1 to enable CUDA support

The commands to install hdf5 (and zlib) development libraries on some popular distributions :

On Debian/Ubuntu : sudo apt-get install libhdf5-dev zlib1g-dev
On Fedora/CentOS : sudo dnf/yum install hdf5-devel zlib-devel
On Arch Linux: sudo pacman -S hdf5
On OS X : brew install hdf5

Other building options

• Building from the Github repository additionally requires invoking autoreconf --install to generate the configure script. autoreconf can be installed on Ubuntu using sudo apt-get install autoconf automake.
• If you want only S/BLOW5 support you can disable HDF5 by invoking ./configure --disable-hdf5 && make.
• If you skip scripts/install-hts.sh and ./configure, hdf5 will be compiled locally. It is a good option if you cannot install hdf5 library system wide. However, building hdf5 takes ages.
• f5c from version 0.8.0 onwards by default requires vector instructions (SSSE3 or higher for Intel/AMD and neon for ARM) for builtin slow5lib. If your processor is an ancient processor with no such vector instructions, invoke make as make no_simd=1.
• You can optionally enable zstd support for builtin slow5lib when building f5c by invoking make zstd=1. This requires zstd 1.3 development libraries installed on your system (libzstd1-dev package for apt, libzstd-devel for yum/dnf and zstd for homebrew).
• On Mac M1 or in any system if ./configure cannot find the hdf5 libraries installed through the package manager, you can specify the location as LDFLAGS=-L/path/to/shared/lib/ CPPFLAGS=-I/path/to/headers/. For example on Mac M1:

   ./configure LDFLAGS=-L/opt/homebrew/lib/ CPPFLAGS=-I/opt/homebrew/include/
   make
  

• Instructions to build a docker image and conda installation are detailed here.
• Other uncommon building options are detailed here.
• An SIMD accelerated version contributed by @dkhyland is available in the simd branch.

NVIDIA CUDA support

To build for the GPU, you need to have the CUDA toolkit installed. Make nvcc (NVIDIA C Compiler) is in your PATH.

The building instructions are the same as above except that you should call make as :

make cuda=1

Optionally you can provide the CUDA architecture as :

make cuda=1 CUDA_ARCH=-arch=sm_xy

If your CUDA library is not in the default location /usr/local/cuda/lib64, point to the correct location as:

make cuda=1 CUDA_LIB=/path/to/cuda/library/

Visit here for troubleshooting CUDA related problems.

Usage

f5c index -d [fast5_folder] [read.fastq|fasta] # or f5c index --slow5 [slow5_file] [read.fastq|fasta]
f5c call-methylation -b [reads.sorted.bam] -g [ref.fa] -r [reads.fastq|fasta] > [meth.tsv] #specify --slow5 [slow5_file] to use a slow5 file instead of fast5
f5c meth-freq -i [meth.tsv] > [freq.tsv]
f5c eventalign -b [reads.sorted.bam] -g [ref.fa] -r [reads.fastq|fasta] > [events.tsv]    #specify --rna for direct RNA data, --slow5 [slow5_file] to use a slow5 file, --pore r10 for R10.4.1 data

Visit the man page for all the commands and options.

Example

Follow the same steps as in Nanopolish tutorial while replacing nanopolish with f5c. If you only want to perform a quick test of f5c :

#download and extract the dataset including sorted alignments
wget -O f5c_na12878_test.tgz "https://f5c.page.link/f5c_na12878_test"
tar xf f5c_na12878_test.tgz

###### Using FAST5 as input ######

#index, call methylation and get methylation frequencies
f5c index -d chr22_meth_example/fast5_files chr22_meth_example/reads.fastq
f5c call-methylation -b chr22_meth_example/reads.sorted.bam -g chr22_meth_example/humangenome.fa -r chr22_meth_example/reads.fastq > chr22_meth_example/result.tsv
f5c meth-freq -i chr22_meth_example/result.tsv > chr22_meth_example/freq.tsv
#event alignment
f5c eventalign -b chr22_meth_example/reads.sorted.bam -g chr22_meth_example/humangenome.fa -r chr22_meth_example/reads.fastq > chr22_meth_example/events.tsv

###### Using SLOW5 as input ######

#index, call methylation and get methylation frequencies
f5c index --slow5 chr22_meth_example/reads.blow5 chr22_meth_example/reads.fastq
f5c call-methylation --slow5 chr22_meth_example/reads.blow5 -b chr22_meth_example/reads.sorted.bam -g chr22_meth_example/humangenome.fa -r chr22_meth_example/reads.fastq > chr22_meth_example/result.tsv
f5c meth-freq -i chr22_meth_example/result.tsv > chr22_meth_example/freq.tsv
#event alignment
f5c eventalign --slow5 chr22_meth_example/reads.blow5 -b chr22_meth_example/reads.sorted.bam -g chr22_meth_example/humangenome.fa -r chr22_meth_example/reads.fastq > chr22_meth_example/events.tsv

Acknowledgement

This reuses code and methods from Nanopolish. The event detection code is from Oxford Nanopore's Scrappie basecaller. Some code snippets have been taken from Minimap2 and Samtools.