m6a RNA科研设计、数据分析、结果解读 精品培训班 实操环节

比对是最耗资源和时间的一步

三、开始比对测序数据，回帖基因组 （reads要回家）

1、传统的hisat2

A - 比对

work_dir=/home4/sjshen/project/m6a/GSE29714_Cell
fastq_cut_dir=$work_dir/0.2_fastq_cut/mouse
hisat2_dir=$work_dir/1.1_hisat2/mouse
hisat2_index=/home/share/hisat2_index

echo $fastq_cut_dir
cd $fastq_cut_dir

species=mm10_tran
thread=5

#for Single-end data
for i in *_cut.fastq.gz
do
echo $i

(nohup hisat2 -p $thread  --dta-cufflinks --no-discordant --no-mixed -t -x $hisat2_index/new_index/$species -U $i -S $hisat2_dir/${i%%_cut.fastq.gz}.hisat2.sam --no-unal > $hisat2_dir/${i%%_cut.fastq.gz}.hisat2.log) &

done

echo $hisat2_dir
cd $hisat2_dir

Tips1:hisat2有官方提供的比对基因组，我们也可以自己建立，需要的同学可以答疑时咨询我们老师
Tips2:hisat2默认输出一条/一对reads的最佳5处匹配位置

#for Pair-end data
for i in *_1_cut.fastq.gz 
do
echo $i
t=${i/_1_cut.fastq.gz/_2_cut.fastq.gz}  
echo $t

(nohup hisat2 -p $thread --dta-cufflinks --no-discordant --no-mixed -t -x $hisat2_index/new_index/$species -1 $i -2 $t -S $hisat2_dir/${i%%_1_cut.fastq.gz}.hisat2.sam --no-unal > $hisat2_dir/${i%%_1_cut.fastq.gz}.hisat2.log) &

done

B - 转换格式保存

work_dir=/home4/sjshen/project/m6a/GSE29714_Cell
hisat2_dir=$work_dir/1.1_hisat2

echo $hisat2_dir
cd $hisat2_dir
mkdir position_sorted_bam
#--------------------------------------------------------------------
#transfer sam to position_sorted_bam
for i in *.sam
do
echo $i
( nohup samtools sort -@ 5 -o ./position_sorted_bam/${i%%.sam}.position_sorted.bam $i ) &
done
#--------------------------------------------------------------------
#build bam index for position_sorted_bam
for i in ./position_sorted_bam/*.bam
do
( nohup samtools index $i ) &
done

C - 简单统计一下

work_dir=/home4/sjshen/project/m6a/GSE29714_Cell
hisat2_dir=$work_dir/1.1_hisat2
echo $hisat2_dir
cd $hisat2_dir

echo -e 'Sample\ttotal_reads(million)\taligned exactly 1 time\tpercentage\taligned >1 times\tpercentage\ttotal_percentage' > hisat2_align_summary.tab

for i in *.log
do
echo $i
awk -v sample=${i%%.hisat2.log} 'BEGIN{FS="[ \t();%]+" ; OFS="\t"}  NR==5 {t=$2} NR==7 {a=$2;b=$3} NR==8 {c=$2;d=$3} END{print sample,t/1e6,a/1e6,b"%",c/1e6,d"%",(b+d)"%"}' $i >> hisat2_align_summary.tab
done

 
《=== hisat2比对结果统计示例 ===》

Tips: hisat2输出的SAM文件中 'NHN' The number of mapped locations for the read or the pair.

D - bam to bigwig

bam_dir=$hisat2_dir/position_sorted_bam
deeptools_bw_dir=$bam_dir/deeptools_bw
mkdir $deeptools_bw_dir
cd $deeptools_bw_dir

for i in $bam_dir/*.position_sorted.bam
do
echo $i
t=${i##*/}
name=${t%%.position_sorted.bam}

nohup bamCoverage -p 10 -b $i -o ${name}.bw  --normalizeUsing RPKM --binSize 10 --smoothLength 30 > ./${name}.bw.log &

done

Tips1:deeptools是NGS比对后分析的四大金刚之一（samtools/bedtools/deeptools/bamtools），后面还会用到他
Tips2:bamCoverage是deeptools工具套件的其中一个函数，产固定步长的bigwig效果很好，他有一个姐妹函数bamCompare后面会讲

B字三剑客 “bam、bigwig、bed” ，三人足以横扫天下，后面会介绍

E - IGV 展示 （准备bw文件）

2、无敌的bwa (不要迷恋哥,哥只是个传说)

A - 比对

work_dir=/home4/sjshen/project/m6a/GSE29714_Cell
fastq_cut_dir=$work_dir/0.2_fastq_cut/mouse
bwa_dir=$work_dir/1.2_bwa/mouse

echo $fastq_cut_dir
cd $fastq_cut_dir

species=/home/share/bwa_index/mm10.fa
thread=5

#for Single-end data
for i in *_cut.fastq.gz
do 
echo $i

( nohup bwa mem -t $thread  $species  $i -M -R '@RG\tID:sample\tSM:sample' -o $bwa_dir/${i%%_cut.fastq.gz}.bwa.sam > $bwa_dir/${i%%_cut.fastq.gz}.bwa.log 2>&1 ) &
done

echo $bwa_dir
cd $bwa_dir

Tips1:bwa需要自己建立比对基因组，需要的同学可以答疑时咨询我们老师
Tips2:bwa不论比对质量都会输出，需要后续筛选

#for Pair-end data
for i in *1_cut.fastq.gz
do 
echo $i
t=${i/1_cut.fastq.gz/2_cut.fastq.gz}

( nohup bwa mem -t $thread  $species  $i $t -M -R '@RG\tID:sample\tSM:sample' -o $bwa_dir/${i%%_1_cut.fastq.gz}.bwa.sam > $bwa_dir/${i%%_1_cut.fastq.gz}.bwa.log 2>&1 ) &

done

B - 转换格式保存，初步筛选数据

## 1.sam to bam ##
for i in *.sam
do
echo $i
( nohup samtools  sort -@ 5 -o ${i%%.sam}.position_sorted.bam $i ) &
done
#--------------------------------------------------------------------
## 2.filter ## 
MAPQ=20
flag=2   #only for pair-end sequencing

for i in *.position_sorted.bam
do
echo $i
nohup samtools view -@ 10 -hb  -f $flag -q $MAPQ $i -o ${i%%.bam}.MAPQ${MAPQ}.bam &
done
#--------------------------------------------------------------------
## 3.index ## 
for i in *MAPQ*.bam
do
echo $i
mv  $i  ${i%%.position_sorted*}.final.bam 
done

for i in *.final.bam 
do
( nohup samtools index $i ) &
done

C - 简单统计一下

work_dir=/home4/sjshen/project/m6a/GSE29714_Cell
fastq_cut_dir=$work_dir/0.2_fastq_cut/mouse
bwa_dir=$work_dir/1.2_bwa/mouse
mkdir $bwa_dir/flagstat
#--------------------------------------------------------------------
# step1- create flagstat from bam files
for i in *.bam
do
echo $i
samtools flagstat -@ 10 $i > ./flagstat/${i/.bam/.flagstat} &
done
#--------------------------------------------------------------------
# step2 - summary from flagstat files
cd $bwa_dir/flagstat
echo -e 'Sample\tTotal mapped fragments(M)\tfilter_keep(M)\tpercentage\tduplicate(M)\tpercentage\tfinal_keep(M)\tpercentage' > flag_state.tab

for i in *.bwa.position_sorted.flagstat
do
name=${i%%.bwa.position_sorted.flagstat}
echo $name
awk -v name=$name '
BEGIN{FS="[ \t();%]+";OFS="\t"}
FILENAME==name".bwa.position_sorted.flagstat" && FNR==5 {total=t=($1/2)/1e6}
FILENAME==name".bwa.position_sorted.MAPQ20.flagstat" && FNR==5 {filter_keep=t=($1/2)/1e6;filter_keep_p=filter_keep/total*100}
END{print name,total,filter_keep,filter_keep_p"%"}' ${name}.bwa.position_sorted.flagstat ${name}.bwa.position_sorted.MAPQ20.flagstat \
>> bwa_align_summary.tab
done

 
《=== bwa比对结果统计示例 ===》

D - bam to bigwig

# with duplicate
for i in ./*.position_sorted.MAPQ20.bam
do
echo $i
t=${i##*/}
name=${t%%.position_sorted*}
nohup bamCoverage  -p 10 -b $i -o ${name}.withdup.bw  --normalizeUsing RPKM --binSize 10 --smoothLength 30 > ./${name}.withdup.bw.log &
done
#--------------------------------------------------------------------
# without duplicate
for i in ./*.final.bam
do
echo $i
t=${i##*/}
name=${t%%.final.bam}
nohup bamCoverage  -p 10 -b $i -o ${name}.rmdup.bw  --normalizeUsing RPKM --binSize 10 --smoothLength 30 > ./${name}.rmdup.bw.log &
done

3、duplicate怎么办?（你还在用picard吗?）

# step1 - mark
for i in *position_sorted.bam
do
echo $i
sambamba markdup -t 10 $i  ${i/.bam/.mark.bam} > ${i/.bam/.mark.log} 2>&1  &
done
#--------------------------------------------------------------------
# step2 - remove
flag=1024
for i in *mark.bam
do
echo $i
nohup samtools view -@ 10 -Shb -F $flag  $i -o ${i%%.position_sorted*}.final.bam &
done

为什么选择sambamba？因为他长得像samtools；
为什么选择sambamba？因为他比picard好用，运行稳定；
为什么选择sambamba？因为他和picard一样结果，通吃SE/PE。
。。。还问？不用问了，用就对了！
不过picard的统计还是很丰富的，有时还是有点作用的

那到底去不去？两个门派都有，哪个好用就入哪个门派吧

Dependencies

• Softwares

  • HISAT2
  • BWA
  • samtools
  • deeptools

• Other

  • sam文件格式
  • STAR
  • TopHat
  • Bowtie2