Pacasus

Tool for detecting and cleaning PacBio / Nanopore long reads after whole genome amplification. Check the poster from the Revolutionizing Next-Generation Sequencing (2nd edition) conference in the source folder: https://github.com/swarris/Pacasus/blob/master/vib2017.pdf.

The BMC Genomics paper https://bmcgenomics.biomedcentral.com/articles/10.1186/s12864-018-5164-1 Correcting palindromes in long reads after whole-genome amplification Warris, Sven; Schijlen, Elio; van de Geest, Henri; Vegesna, Rahulsimham; Hesselink, Thamara; te Lintel Hekkert, Bas; Sanchez Perez, Gabino; Medvedev, Paul; Makova, Kateryna D.; de Ridder, Dick; 2018/11/06

It uses the pyPaSWAS framework for sequence alignment (https://github.com/swarris/pyPaSWAS)

Platforms supported:

• NVIDIA GPU using CUDA (compute capability 1.3 and higher)
• NVIDIA GPU using OpenCL
• Intel CPU using OpenCL
• Intel Xeon Phi accelerator using OpenCL
• Other systems supporting OpenCL (AMD, Intel GPUs, etc) should be able to run the software, but are untested.

Installation

In most cases it is enough to clone the repository.

git clone https://github.com/swarris/Pacasus.git

After that, you need to install:

• pip (https://docs.python.org/2.7/installing/)
• numpy: sudo pip install numpy (or pip install --user numpy)
• BioPython: sudo pip install Biopython (or pip install --user Biopython)
• (for Python 2 only) ConfigParser: sudo pip install ConfigParser (or pip install --user ConfigParser)
• In some cases, the python development packages are required (Ubuntu: sudo apt-get install python-dev)

Making use of the CUDA version (also recommended when using the OpenCL version on a NVIDIA GPU):

• Download CUDA sdk: https://developer.nvidia.com/cuda-downloads
• sudo pip install pyCuda (http://mathema.tician.de/software/pycuda/)

Making use of the OpenCL version:

• check dependencies or downloads for your system. See this wiki for some great pointers: http://wiki.tiker.net/OpenCLHowTo
• sudo pip install pyOpenCL

Getting pyPaSWAS: pyPaSWAS is required as a module. Run the following two commands in the Pacasus root folder:

• git submodule init
• git submodule update

Running the software

The read file is mandatory, by setting the options one can specify the type of input file (default: fasta), the output file and a log file. When requested, Pacasus will stop if the output file exists.

Run it by calling:

• python pacasus.py |options| readsFile

Help file:

• python pacasus.py --help

Selection your device

By default, pacasus will use the first CPU device. This can be changed by using:

• --device_type=[CPU|GPU]
• --platform_name=[Intel|NVIDIA]
• --framework=[opencl|CUDA]
• --device=[int]

For example, this will select the CPU: --device_type=CPU --platform_name=Intel --framework=opencl

This will select the second NVIDIA GPU: --device_type=GPU --platform_name=NVIDIA --framework=CUDA --device=1

Example

Use a fasta-file:

• python pacasus.py reads.fasta -o cleaned.fasta --loglevel=DEBUG

Key settings for cleaning

The following settings impact the number of reads which will be cleaned. These settings define the minimal properties the alignment should have to be identified as a palindrome sequence:

• --filter_factor=0.01
• --query_coverage=0.01
• --query_identity=0.01
• --relative_score=0.01
• --base_score=1.0

These settings are relatively strict. Are your reads very noisy, only a few reads are cleaned or the de novo is still very fragmented? Then lower these values to for example:

• --filter_factor=0.00001
• --query_coverage=0.0001
• --query_identity=0.0001
• --relative_score=0.0001
• --base_score=0.25

In those cases it best to also increase the minimum read length after cleaning, otherwise the output will be full of (very) short sequences (--minimum_read_length=500).

Table 1. Key command line options

Option	Long version	Description
-h	--help	This help
-L	--logfile	Path to the log file
	--loglevel	Specify the log level for log file output. Valid options are DEBUG, INFO, WARNING, ERROR and CRITICAL
-o	--output	Path to the output file. Default ./output
-O	--overrideOutput	When output file exists, override it (T/F). Default T (true)
-1	--filetype1	File type of the first input file. See bioPython IO for available options. Default fasta
-G		Float value for the gap penalty. Default -5
-q		Float value for a mismatch. Default -3
-r		Float value for a match. Default 1
	--any	Float value for an ambiguous nucleotide. Default 1
	--other	Float value for an ambiguous nucleotide. Default 1
	--device	Integer value indicating the device to use. Default 0 for the first device.
-c		Option followed by the name of a configuration file. This option allows for short command line calls and administration of used settings.

If you have questions, please e-mail s.warris@gmail.com