Modified by wym6912

Split the consensus sequence and add static linking program.

Compile with static linking

make STATIC_LINK=1

Contact

wym6912 (wym6912@outlook.com)

TSTA

pairwise and multiple sequence alignment accelerated by simd and threads

Introduction

TSTA is a sequence alignment algorithm that combines SIMD instruction sets and thread acceleration. It uses the difference method, stripe method, and anti-diagonal method, and uses partial order alignment (POA) in multi sequence alignment.

TSTA is committed to achieving faster comparison speed. TSTA only supports global alignment and allows linear and affine gap penalties. It supports SSE4.2/AVX2/AVX512 vectorization, with different choices based on hardware support.

Installation

git clone https://github.com/bxskdh/TSTA.git
cd TSTA
make

Usage and run

Pairwise sequence alignment

./TSTA_psa -1 ./example/psa/seq/seqa1.fa -2 ./example/psa/seq/seqb1.fa
# maxsorce=-5

no-backtracing PSA

./TSTA_psa_notrace -1 ./example/psa/seq/seqa1.fa -2 ./example/psa/seq/seqb1.fa
# maxsorce=-5

Multiple sequence alignment

./TSTA_msa -i ./example/msa/seq/seq1.fa
# seq:5000, seqN:5120, poa:5000
# poa_len=5000 ,seq_len=5000 ,trace_sub:[num1]=4999 [num2]=4999 ,lastsorce=-5451
# poa_add_len:1885
# seq:5000, seqN:5120, poa:6885
# poa_len=6885 ,seq_len=5000 ,trace_sub:[num1]=6884 [num2]=4999 ,lastsorce=-3101
# poa_add_len:1714
# seq:5000, seqN:5120, poa:8599
# poa_len=8599 ,seq_len=5000 ,trace_sub:[num1]=8598 [num2]=4999 ,lastsorce=-1776
# poa_add_len:1531
# seq:5000, seqN:5120, poa:10130
# poa_len=10130 ,seq_len=5000 ,trace_sub:[num1]=10129 [num2]=4999 ,lastsorce=-870
# poa_add_len:1338

Tip

While you are running the program in msa mode, we recommend that the sequences used for alignment should be less than 200,000bp and more than 50,000bp. If the sequences are too long, the program may run out of memory.

Operating system

The Linux platform is supported by TSTA.It run and tested on CentOS Linux 7.

Contact

• Peiyu Zong peiyuzong8@gmail.com (Designed and implemented core algorithms)
• Wenpeng Deng 1732889554@qq.com (Optimized some of the code)