ParaPindel

This is a multi-processes parallel version of Pindel to accelerate structural variation detection. Pindel's implementation can be found at https://github.com/xjtu-omics/pindel

Publication

Yang, Y., Wang, X., Xu, Y., Yang, C., Jiang, B., & Peng, S. (2021, December). ParaPindel: a scalable coordinated parallel detection framework for human genome-wide structural variation. In 2021 IEEE International Conference on Bioinformatics and Biomedicine (BIBM) (pp. 574-579). IEEE.

To run ParaPindel, follow these steps:

1. Configure MPI environment

I recommend MPICH or intelMPI

2. Download ParaPindel

git clone https://github.com/pengsl-lab/ParaPindel.git

3. Compile ParaPindel using the commands

cd ParaPindel
./INSTALL ./htslib

If you have installed the HTSLIB library, use the following command:

./INSTALL /home/user/path/to/your/htslib

Run ParaPindel

(1) Create a new directory to store the results

mkdir result

(2) Create a configuration file test.config for locating the bam file, including the location of the bam file (it is best to use an absolute path), InsertSize and a field named for the bam file. For example:

/home/user/path/to/yuor/test.bam 500 test

(3) If it is on a workstation with a single node, use the following command to run ParaPindel

mpirun -np 4 ./paraPindel -f /path/to/your/reference.fasta -i /path/to/your/test.config -w 5 -W 5 -c ALL -T 8 -o ./result/test

The -np parameter represents the number of processes. Other parameters can be viewed using the ./paraPindel -h command.

(4) If it is on a multi-node cluster with slurm, use the following command（The specific submission instructions depend on the cluster you are using）:

srun -N 4 ./paraPindel -f /path/to/your/reference.fasta -i /path/to/your/test.config -w 5 -W 5 -c ALL -T 8 -o ./result/test

Or submit according to the job method with the following command:

sbatch -N 4 ./paraPindel -f /path/to/your/reference.fasta -i /path/to/your/test.config -w 5 -W 5 -c ALL -T 8 -o ./result/test

-np is a parameter that needs to be added when running the mpirun command, not a parameter of ./paraPindel, which represents how many processes are used for parallel detection. Similarly, -N is the parameter when running the srun or sbatch command.
The -T parameter is implemented in the current Pindel version (https://github.com/xjtu-omics/pindel) and represents the number of threads. -T 8 is specified here, which means that 8 threads are used in each process.

Combine the calculation results of each process, take deletion variation as an example.

cd ./result
cat test_D_* > test_D
rm test_D_*

Convert the result into vcf format if the reference is hg19.

../pindel2vcf -p test_D -r /path/to/your/reference.fasta -R hg19 -d 20210606 -G -v test_D.vcf
cat test_D.vcf | vcf-sort > test_D_sorted.vcf

The pindel2vcf tool is used to convert the variation results into vcf format, which is implemented in Pindel. For the specific usage of pindel2vcf, please refer to the Pindel homepage (https://github.com/xjtu-omics/pindel).

If you have any questions or comments, please feel free to email:[email protected]