PEATH/G is a GPGPU version of PEATH, a novel SIH algorithm based on the estimation of distribution algorithm (EDA).
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.
After downloading .cu code, complie the source code using NVCC compiler
nvcc -o PEATH-G PEATH-G.cu -arch=sm_35 -rdc=true -lcudadevrt
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
We used the data set for testing the performance(running time) of our implementation
- Fosmid dataset (Duitama et al. 2012) which has been widely used to assess and compare SIH algorithms.