PEATH/G: Fast Single-Individual Haplotyping method using GPGPU

PEATH/G is a GPGPU version of PEATH, a novel SIH algorithm based on the estimation of distribution algorithm (EDA).

System Environment

We have tested for compiling and running the code in conventional desktop computer as follows:

• Hardware: Intel Core i5, 16GB RAM, and NVIDIA GeForce GTX950 graphics card.
• OS: Windows 10 64-bits
• IDE: Visual Studio 2017 integrated with CUDA toolkit 10.1

Our program works well also in other environment such as Ubuntu Linux 64-bits with GTX 1050 or GTX 1660Ti graphic card.

Compiling PEATH/G

After downloading .cu code, complie the source code using NVCC compiler

nvcc -o PEATH-G PEATH-G.cu -arch=sm_35 -rdc=true -lcudadevrt

Running PEATH/G

To run PEATH/G, use the following command:

./PEATH-G <input_file> <output_file> (param)

<input_file> is an input matrix for sequence reads and <output_file> contains phased haplotype.

(param) is an optional parameter for time/accuracy tradeoff which is a positive integer (default: 50).

ex) ./PEATH-G chr1.matrix.SORTED chr1.haplo

Data Set for testing

We used the data set for testing the performance(running time) of our implementation

1. Fosmid dataset (Duitama et al. 2012) which has been widely used to assess and compare SIH algorithms.