maxbaird / masa-core

Multi-Platform Architecture for Sequence Aligner (MASA) is a flexible software framework that simplifies the creation of DNA sequence alignment applications in multiple hardware/software platforms. This framework is based on CUDAlign and it supports huge DNA sequences with more than 200 million base pairs.

MASA-Core

Multi-Platform Architecture for Sequence Aligner (MASA) is a flexible software framework that simplifies the creation of DNA sequence alignment applications in multiple hardware/software platforms. This framework supports the alignment of huge DNA sequences with more than 200 million base pairs (MBP).

The MASA-Core library is based on CUDAlign, a tool able to align huge sequences in CUDA capable GPUs. The MASA-Core contains 90% of the original CUDAlign source code, isolating the platform-independent features. Thus, any developer may create "aligners" for different hardware/software platforms, reimplementing only some methods of the original code and linking it against the portable MASA-Core library. Furthermore, new platform-independent optimizations may be deployed into the MASA-Core, leading to a wider contribution to all the specific aligners implementations. Platform-specific aligners will be maintained in separate repositories, with a static copy of the masa-core source.

Main Features:

The MASA-Core contains the following features:

• Produces optimal alignments for DNA sequences of any length (even greater than 200 MBP)
• Based on Smith-Waterman and Needleman-Wunsch algorithms
• Uses Myers-Miller strategy during the matrix computation (linear memory complexity)^IPDPS2011
• Produces Local, Semi-Global and Global optimal alignments
• Multi-node support for homogeneous^CCGRID2014 and heterogeneous hardware^PPOPP2014
• Block Pruning optimization^TPDS2012
• Portable C/C++ code. Compiled in Linux and MacOS X, for 32 and 64 bit platforms

Download

Latest Version: masa-core-1.3.9.1024.tar.gz

Compiling the static library (libmasa.a)

tar -xvzf masa-core-1.3.9.1024.tar.gz
cd masa-core-1.3.9.1024
./configure
make

To create MASA executable files, download one of the extensions listed below. Each extensions project contains a copy of the MASA-Core source.

Supported Platforms:

Each project that supplies a different Aligner linked with MASA-Core is called a MASA-Extension. Here is a list with some MASA-Extensions for some supported platforms.

• MASA-CUDAlign: Extension that uses CUDA NVidia GPUs (based on the original CUDAlign code)
• MASA-OpenMP: Uses OpenMP to computes many blocks in parallel using CPU or Intel Phi cores
• MASA-OmpSs: Uses OmpSs to run parallel tasks respecting dependencies
• MASA-Serial: This is a simple extension that runs with serial CPU code

Executing a MASA extension

./masa-extension [options] seq1.fasta seq2.fasta

All the command line arguments can be retrieved using the --help parameter. See the wiki for a list of command line examples.

Documentation

See the doxygen pages.

License:

MASA-Core is an open source project with public license (GPLv3). A copy of the LICENSE is maintained in this repository.

---

References:

[TOPC2016]	MASA: a Multi-Platform Architecture for Sequence Aligners with Block Pruning. TOPC:2(4):28. Edans Sandes, Guillermo Miranda, Xavier Martorell, Eduard Ayguadé, George Teodoro, Alba Melo.
[TPDS2016]	Genome Wide Alignment in GPU Cluster with Incremental Speculative Traceback. TPDS 2016. Edans Sandes, Guillermo Miranda, Alba Melo, Xavier Martorell, Eduard Ayguadé.
[CCGRID2014]	CUDAlign 3.0: Parallel Biological Sequence Comparison in Large GPU Clusters. CCGRID 2014:160-169. Edans Sandes, Guillermo Miranda, Alba Melo, Xavier Martorell, Eduard Ayguadé.
[PPOPP2014]	Fine-grain parallel megabase sequence comparison with multiple heterogeneous GPUs. PPOPP 2014:383-384. Edans Sandes, Guillermo Miranda, Alba Melo, Xavier Martorell, Eduard Ayguadé
[TPDS2013]	Retrieving Smith-Waterman Alignments with Optimizations for Megabase Biological Sequences using GPU. TPDS:24:5:1009-1021. Edans Sandes, Alba Melo
[IPDPS2011]	Smith-Waterman Alignment of Huge Sequences with GPU in Linear Space. IPDPS 2011: 1199-1211. Edans Sandes, Alba Melo
[PPOPP2010]	CUDAlign: using GPU to accelerate the comparison of megabase genomic sequences. PPOPP 2010: 137-146. Edans Sandes, Alba Melo