human_genomics_pipeline是一个基于bwa、gatk4的基因组Snakemake分析流程，可以用来处理双端测序数据。整个流程是从fastq处理到最后的vcf，可以利用GPU加速（NVIDIA GPU）。在配置文件中进行简单设置，指定特定物种的参考基因组和变异数据库来完成基因组的数据分析。

1）单个样本处理流程：

非GPU处理流程
GPU处理流程

2）队列数据处理流程：

非GPU处理流程
GPU处理流程

3）系统需求：

4）数据示例：

注意：输入双端测序fastq文件需要使用_1和_2标识，其他注意事项可以参考：访问

5）GATK resource bundle下载

参考：https://github.com/bahlolab/bioinfotools/blob/master/GATK/resource_bundle.md

6）配置文件（编辑config.yaml文件）


##############################
###### Overall workflow ######
##############################

# Type of input data (either 'Single' or 'Cohort')
DATA: ""

# Should the pipeline be GPU accelerated where possible? (either 'Yes' or 'No')
GPU_ACCELERATED: ""

# File path to the reference genome (.fasta)
REFGENOME: ""

# File path to dbSNP database
dbSNP: ""

# Temporary file directory
TEMPDIR: ""

# Whole exome sequence settings (leave blank if analysing other data such as whole genome sequence data)
WES:
  # File path to the exome capture regions over which to operate (prefix with the '-L' flag)
  INTERVALS: ""
  # Padding (in bp) to add to each region (prefix with the '-ip' flag)
  PADDING: ""

##############################
##### Pipeline resources #####
##############################

# Number of threads to use per rule/sample for multithreaded rules, multithreading will significantly speed up these rules (diminishing speed gains beyond 8 threads)
THREADS:

# Maximum memory usage per rule/sample (eg. '40g' for 40 gigabytes, this should suffice for exomes)
MAXMEMORY: ""

# Maximum number of GPU's to be used per rule/sample for gpu-accelerated runs (eg `1` for 1 GPU)
GPU:

##############################
########## Trimming ##########
##############################

# Whether or not to trim the raw fastq reads (either 'Yes' or 'No')
TRIM: ""

# If trimming, choose the adapter sequence to be trimmed (eg. `--illumina`, `--nextera` or `--small_rna`)
TRIMMING:
  ADAPTERS: ""

##############################
##### Base recalibration #####
##############################

# Resources to used for base recalibration (prefix each resource with the '--known-sites' flag if not gpu accelerated and the '--knownSites' if gpu accelerated)
RECALIBRATION:
  RESOURCES: ""

HPC配置（cluster.json文件，可选）：


{
    "__default__" :
    {
        "account" : "",
        "partition" : ""
    }
}

如果是使用是使用slurm，可以参考：如何在snakemake设置slurm

7）运行流程

当配置好配置文件就可以在bash下运行流程

bash dryrun.sh或者bash run.sh


# dryrun.sh内容
snakemake \
--dryrun \
--cores 32 \
--resources mem_mb=150000 \
--resources gpu=2 \
--use-conda \
--conda-frontend mamba \
--configfile ../config/config.yaml

# run.sh内容
snakemake \
--cores 32 \
--resources mem_mb=150000 \
--resources gpu=2 \
--use-conda \
--configfile ../config/config.yaml

整个流程就介绍到此了，本流程的亮点是GPU的运用值得学习和借鉴。

参考资料：

1.https://github.com/ESR-NZ/human_genomics_pipeline

阅读: 784