首先maf格式的somatic突变数据制作成为annovar软件的输入格式：
cut -f 5-7,12,13,1,16 human_brca_all_mutect2.maf |cut -f 2-7 > 1
cut -f 5-7,12,13,1,16 human_brca_all_mutect2.maf |cut -f 1 > 2
paste 1 2 > for_annovar.input ### 共 13027 位点然后运行annovar软件的批量注释功能
bin=/home/haitaowang/Database/hg38/Annovar/
db=/home/haitaowang/Database/hg38/Annovar/
perl $bin/table_annovar.pl for_annovar.input $db -buildver hg38 -out tmp \
-protocol refGene,cytoBand,esp6500siv2_all,exac03,gnomad_genome,cosmic70,1000g2015aug_all,clinvar_20170905 \
-operation g,r,r,f,f,f,f,f -nastring NA因为数据库较多，所以注释耗时很长，注释结果如下：
551K Aug 17 16:52 for_annovar.input
99K Aug 17 17:10 tmp.hg38_ALL.sites.2015_08_dropped
423K Aug 17 17:10 tmp.hg38_ALL.sites.2015_08_filtered
459K Aug 17 16:57 tmp.hg38_avsnp150_dropped
174K Aug 17 16:57 tmp.hg38_avsnp150_filtered
56K Aug 17 17:11 tmp.hg38_clinvar_20170905_dropped
477K Aug 17 17:11 tmp.hg38_clinvar_20170905_filtered
35K Aug 17 17:10 tmp.hg38_cosmic70_dropped
470K Aug 17 17:10 tmp.hg38_cosmic70_filtered
667K Aug 17 16:53 tmp.hg38_cytoBand
366K Aug 17 17:10 tmp.hg38_dbnsfp33a_dropped
445K Aug 17 17:10 tmp.hg38_dbnsfp33a_filtered
115K Aug 17 16:53 tmp.hg38_esp6500siv2_all_dropped
410K Aug 17 16:53 tmp.hg38_esp6500siv2_all_filtered
423K Aug 17 16:53 tmp.hg38_exac03_dropped
307K Aug 17 16:53 tmp.hg38_exac03_filtered
286K Aug 17 16:53 tmp.hg38_genomicSuperDups
731K Aug 17 16:55 tmp.hg38_gnomad_genome_dropped
178K Aug 17 16:55 tmp.hg38_gnomad_genome_filtered
153K Aug 17 17:11 tmp.hg38_intervar_20180118_dropped
436K Aug 17 17:11 tmp.hg38_intervar_20180118_filtered
40K Aug 17 17:12 tmp.hg38_mcap_dropped
457K Aug 17 17:12 tmp.hg38_mcap_filtered
6.3M Aug 17 17:12 tmp.hg38_multianno.txt
48K Aug 17 17:12 tmp.hg38_revel_dropped
447K Aug 17 17:12 tmp.hg38_revel_filtered
68K Aug 17 17:12 tmp.invalid_input
956 Aug 17 17:12 tmp.log
208K Aug 17 16:53 tmp.refGene.exonic_variant_function
68K Aug 17 16:53 tmp.refGene.invalid_input
1.2K Aug 17 16:53 tmp.refGene.log
761K Aug 17 16:53 tmp.refGene.variant_function简单数一下：
1454 tmp.hg38_ALL.sites.2015_08_dropped
7400 tmp.hg38_avsnp150_dropped
170 tmp.hg38_clinvar_20170905_dropped
326 tmp.hg38_cosmic70_dropped
937 tmp.hg38_dbnsfp33a_dropped
1800 tmp.hg38_esp6500siv2_all_dropped
4262 tmp.hg38_exac03_dropped
7305 tmp.hg38_gnomad_genome_dropped
1167 tmp.hg38_intervar_20180118_dropped
648 tmp.hg38_mcap_dropped
890 tmp.hg38_revel_dropped需要一个个数据库来解读。
首先，被千人基因组计划的人群频率0.05过滤掉的坐标拿出来：
perl -alne '{print if $F[1]>0.05}' tmp.hg38_ALL.sites.2015_08_dropped > filter_by_1000g.pos有了这些坐标，就可以去过滤我们的原始maf文件啦。
cat filter_by_1000g.pos ../human_brca_all_mutect2.maf |perl -alne '{if(/^1000/){$h{"$F[2]\t$F[3]"}=1}else{print unless exists $h{"$F[4]\t$F[5]"}}}' > filter_by_1000g.maf同理，其它数据库也是同样的操作
perl -alne '{print if (split(",",$F[1]))[0]>0.05}' tmp.hg38_gnomad_genome_dropped > filter_by_gnomad.pos
perl -alne '{print if (split(",",$F[1]))[0]>0.05}' tmp.hg38_exac03_dropped > filter_by_exac03.pos
cat filter_by_exac03.pos filter_by_1000g.maf |perl -alne '{if(/^exac03/){$h{"$F[2]\t$F[3]"}=1}else{print unless exists $h{"$F[4]\t$F[5]"}}}' > filter_by_1000g_exac03.maf
cat filter_by_gnomad.pos filter_by_1000g_exac03.maf|perl -alne '{if(/^gnomad_genome/){$h{"$F[2]\t$F[3]"}=1}else{print unless exists $h{"$F[4]\t$F[5]"}}}' > filter_by_1000g_exac03_gnomeAD.maf过滤效果如下：
wc -l *.maf
9955 filter_by_1000g_exac03_gnomeAD.maf
10588 filter_by_1000g_exac03.maf
12141 filter_by_1000g.maf过滤后的maf作为肿瘤外显子分析结果应该是会更合理。
生信技能树GATK4系列教程
GATK4的gvcf流程
你以为的可能不是你以为的
新鲜出炉的GATK4培训教材全套PPT，赶快下载学习吧
曾老湿最新私已：GATK4实战教程
GATK4的CNV流程-hg38
然后是 CNV相关工具
WES的CNV探究-conifer软件使用
单个样本NGS数据如何做拷贝数变异分析呢
肿瘤配对样本用varscan 做cnv分析
使用cnvkit来对大批量wes样本找cnv
使用sequenza软件判定肿瘤纯度
还有vcf和maf的工具：
安装VEP及其注释数据库
肿瘤突变数据可视化神器-maftools
把vcf文件转换为maf格式，肿瘤外显子上游分析教程到此为止
GATK4的mutect2流程
终于看到了一个完整的mutect2使用脚本
小鼠全基因组数据分析