记录一下环状RNA高通量测序数据分析pipeline

最新推荐文章于 2024-10-10 22:13:52 发布
Finasty
最新推荐文章于 2024-10-10 22:13:52 发布
阅读量360
点赞数
文章标签： linux 服务器数据分析
本文链接：https://blog.csdn.net/Finasty/article/details/123520449
版权
//创建基础文件夹
mkdir /work_dictionary/software
mkdir /work_dictionary/script
mkdir /work_dictionary/index
//安装sratoolkit软件包
wget -c https://ftp-trace.ncbi.nlm.nih.gov/sra/sdk/3.0.0/sratoolkit.3.0.0-centos_linux64.tar.gz
//使用Xftp将circRNA分析软件传输到文件夹
//下载并构建索引
wget -c http://ftp.ensembl.org/pub/release-105/fasta/homo_sapiens/dna/Homo_sapiens.GRCh38.dna_sm.primary_assembly.fa.gz
wget -c http://ftp.ensembl.org/pub/release-105/gtf/homo_sapiens/Homo_sapiens.GRCh38.105.gtf.gz
bwa index -a bwtsw Homo_sapiens.GRCh38.dna_sm.primary_assembly.fa

//创建工作项目文件夹
mkdir /work_dictionary
mkdir /work_dictionary/00.SRR_files
mkdir /work_dictionary/01.reads
mkdir /work_dictionary/02.bwa
mkdir /work_dictionary/03.CIRI2
mkdir /work_dictionary/04.CIRI_AS
mkdir /work_dictionary/05.CIRI_pipeline
mkdir /work_dictionary/06.CIRI_ro1
mkdir /work_dictionary/07.bwa_mem_ro1
mkdir /work_dictionary/08.CIRI_ro2
mkdir /work_dictionary/09.CIRI_merge
mkdir /work_dictionary/10.CIRI_vis

//00.SRR_files
cd mkdir /work_dictionary/00.SRR_files
cp /work_dictionary/script/wget.sh /work_dictionary/00.SRR_files/wget.sh
qsub wget.sh

//01.reads
cd input=/work_dictionary/00.SRR_files
ls -lR |grep SRR|awk '{print $9}' > fastq_samples.list
cp /work_dictionary/script/fastq_dump_sample.sh /work_dictionary/00.SRR_files/fastq_dump_sample.sh
sed -i 's/dictionary/xxx_circseq/g' fastq_dump_sample.sh
for i in `cat fastq_samples.list`; do cp fastq_dump_sample.sh fastq_dump_${i}.sh; done
for i in `cat fastq_samples.list`; do sed -i 's/sample/'$i'/g' fastq_dump_${i}.sh; done
for i in `cat fastq_samples.list`; do qsub fastq_dump_${i}.sh; done

//02.bwa
cd /work_dictionary/01.reads
ls -lR |grep _1.fastq.gz$|awk '{print $9}' > bwa_samples.list //$代表结尾
sed -i 's/_1.fastq.gz//g' bwa_samples.list //用空白替换_1.fastq.gz这样的尾缀，等同于去除尾缀
cp /work_dictionary/script/bwa_sample.sh /work_dictionary/01.reads/bwa_sample.sh //复制脚本到工作目录
sed -i 's/dictionary/xxx_circseq/g' bwa_sample.sh //修改sample脚本中输入和输出的目录为当前工作目录
for i in `cat bwa_samples.list`; do cp bwa_sample.sh bwa_${i}.sh; done //复制sample脚本并批量重命名脚本
for i in `cat bwa_samples.list`; do sed -i 's/sample/'$i'/g' bwa_${i}.sh; done //遍历list中各行，替换sample脚本中的样本名称
for i in `cat bwa_samples.list`; do qsub bwa_${i}.sh; done //批量提交作业

//03.CIRI2
cd /work_dictionary/02.bwa
ls -lR |grep .sam$|awk '{print $9}' > CIRI2_samples.list
sed -i 's/.sam//g' CIRI2_samples.list
cp /work_dictionary/script/CIRI2_sample.sh /work_dictionary/02.bwa/CIRI2_sample.sh
sed -i 's/dictionary/xxx_circseq/g' CIRI2_sample.sh
for i in `cat CIRI2_samples.list`; do cp CIRI2_sample.sh CIRI2_${i}.sh; done
for i in `cat CIRI2_samples.list`; do sed -i 's/sample/'$i'/g' CIRI2_${i}.sh; done
for i in `cat CIRI2_samples.list`; do qsub CIRI2_${i}.sh; done

//04.CIRI_AS
cd /work_dictionary/03.CIRI2
ls -lR |grep .ciri$|awk '{print $9}' > CIRI_AS_samples.list
sed -i 's/.ciri//g' CIRI_AS_samples.list
cp /work_dictionary/script/CIRI_AS_sample.sh /work_dictionary/03.CIRI2/CIRI_AS_sample.sh
sed -i 's/dictionary/xxx_circseq/g' CIRI_AS_sample.sh
for i in `cat CIRI_AS_samples.list`; do cp CIRI_AS_sample.sh CIRI_AS_${i}.sh; done
for i in `cat CIRI_AS_samples.list`; do sed -i 's/sample/'$i'/g' CIRI_AS_${i}.sh; done
for i in `cat CIRI_AS_samples.list`; do qsub CIRI_AS_${i}.sh; done

//05.CIRI_pipeline
cd /work_dictionary/04.CIRI_AS
ls -lR |grep _library_length.list$|awk '{print $9}' > CIRI_pipeline_samples.list
sed -i 's/_library_length.list//g' CIRI_pipeline_samples.list
cp /work_dictionary/script/CIRI_pipeline_sample.sh /work_dictionary/04.CIRI_AS/CIRI_pipeline_sample.sh
sed -i 's/dictionary/xxx_circseq/g' CIRI_pipeline_sample.sh
for i in `cat CIRI_pipeline_samples.list`; do cp CIRI_pipeline_sample.sh CIRI_pipeline_${i}.sh; done
for i in `cat CIRI_pipeline_samples.list`; do sed -i 's/sample/'$i'/g' CIRI_pipeline_${i}.sh; done
for i in `cat CIRI_pipeline_samples.list`; do qsub CIRI_pipeline_${i}.sh; done

//06.CIRI_ro1
cd /work_dictionary/05.CIRI_pipeline
cp /work_dictionary/04.CIRI_AS/CIRI_pipeline_samples.list /work_dictionary/05.CIRI_pipeline/CIRI_ro1_samples.list
cp /work_dictionary/script/CIRI_ro1_sample.sh /work_dictionary/05.CIRI_pipeline/CIRI_ro1_sample.sh
sed -i 's/dictionary/xxx_circseq/g' CIRI_ro1_samples.list
for i in `cat CIRI_ro1_samples.list`; do cp CIRI_ro1_sample.sh CIRI_ro1_${i}.sh; done
for i in `cat CIRI_ro1_samples.list`; do sed -i 's/sample/'$i'/g' CIRI_ro1_${i}.sh; done
for i in `cat CIRI_ro1_samples.list`; do qsub CIRI_ro1_${i}.sh; done

//07.bwa_mem_ro1
cd /work_dictionary/06.CIRI_ro1
ls -lR |grep _ro1.fq$|awk '{print $9}' > bwa_mem_ro1_samples.list
sed -i 's/_ro1.fq//g' bwa_mem_ro1_samples.list
cp /work_dictionary/script/bwa_mem_ro1_sample.sh /work_dictionary/06.CIRI_ro1/bwa_mem_ro1_sample.sh
sed -i 's/dictionary/xxx_circseq/g' bwa_mem_ro1_sample.sh
for i in `cat bwa_mem_ro1_samples.list`; do cp bwa_mem_ro1_sample.sh bwa_mem_ro1_${i}.sh; done
for i in `cat bwa_mem_ro1_samples.list`; do sed -i 's/sample/'$i'/g' bwa_mem_ro1_${i}.sh; done
for i in `cat bwa_mem_ro1_samples.list`; do qsub bwa_mem_ro1_${i}.sh; done

//08.CIRI_ro2
cd /work_dictionary/07.bwa_mem_ro1
ls -lR |grep _ro1.sam$|awk '{print $9}' > CIRI_ro2_samples.list
sed -i 's/_ro1.sam//g' CIRI_ro2_samples.list
cp /work_dictionary/script/CIRI_ro2_sample.sh /work_dictionary/07.bwa_mem_ro1/CIRI_ro2_sample.sh
sed -i 's/dictionary/xxx_circseq/g' CIRI_ro2_sample.sh
for i in `cat CIRI_ro2_samples.list`; do cp CIRI_ro2_sample.sh CIRI_ro2_${i}.sh; done
for i in `cat CIRI_ro2_samples.list`; do sed -i 's/sample/'$i'/g' CIRI_ro2_${i}.sh; done
for i in `cat CIRI_ro2_samples.list`; do qsub CIRI_ro2_${i}.sh; done

//09.CIRI_merge
cd /work_dictionary/08.CIRI_ro2
ls -lR |grep _ro2_info.list$|awk '{print $9}' > CIRI_merge_samples.list
sed -i 's/_ro2_info.list//g' CIRI_merge_samples.list
cp /work_dictionary/script/CIRI_merge_sample.sh /work_dictionary/08.CIRI_ro2/CIRI_merge_sample.sh
sed -i 's/dictionary/xxx_circseq/g' CIRI_merge_sample.sh
for i in `cat CIRI_merge_samples.list`; do cp CIRI_merge_sample.sh CIRI_merge_${i}.sh; done
for i in `cat CIRI_merge_samples.list`; do sed -i 's/sample/'$i'/g' CIRI_merge_${i}.sh; done
for i in `cat CIRI_merge_samples.list`; do qsub CIRI_merge_${i}.sh; done

//10.CIRI_vis
cd /work_dictionary/09.CIRI_merge
ls -lR |grep _merge_circRNA_detail.anno$|awk '{print $9}' > CIRI_vis_samples.list
sed -i 's/_merge_circRNA_detail.anno//g' CIRI_vis_samples.list
cp /work_dictionary/script/CIRI_vis_sample.sh /work_dictionary/09.CIRI_merge/CIRI_vis_sample.sh
sed -i 's/dictionary/xxx_circseq/g' CIRI_vis_sample.sh
for i in `cat CIRI_vis_samples.list`; do cp CIRI_vis_sample.sh CIRI_vis_${i}.sh; done
for i in `cat CIRI_vis_samples.list`; do sed -i 's/sample/'$i'/g' CIRI_vis_${i}.sh; done
for i in `cat CIRI_vis_samples.list`; do qsub CIRI_vis_${i}.sh; done