写在前面

为什么会用这个工具呢
因为我听说很快，并且被 samtools markdup 和 picard 伤到了。用 samtools markdup的时候提醒我要先 fixmate 并且 sort 按照 read name 来，可是我先前是按照默认的sort方式来的，emmm。gatk picard 去除重复后，比原先文件还大，加了什么鬼东西啊

附上此工具链接

http://lomereiter.github.io/sambamba/docs/sambamba-markdup.html

开始

  1. gzip -d sambamba-0.6.8.gz
  2. chmod a+x sambamba-0.6.8
  4. ./sambamba-0.8.6

下载解压，放进环境变量，就是如此简单，不需要安装。

NAME

sambamba-markdup - finding duplicate reads in BAM file

SYNOPSIS

sambamba markdup OPTIONS

DESCRIPTION

Marks (by default) or removes duplicate reads. For determining whether a read is a duplicate or not, the same criteria as in Picard are used.

OPTIONS

-r, --remove-duplicates
remove duplicates instead of just marking them

-t, --nthreads=NTHREADS
number of threads to use

-l, --compression-level=N
specify compression level of the resulting file (from 0 to 9)”);

-p, --show-progress
show progressbar in STDERR

--tmpdir=TMPDIR
specify directory for temporary files; default is /tmp

--hash-table-size=HASHTABLESIZE
size of hash table for finding read pairs (default is 262144 reads); will be rounded down to the nearest power of two; should be > (average coverage) * (insert size) for good performance

--overflow-list-size=OVERFLOWLISTSIZE
size of the overflow list where reads, thrown away from the hash table, get a second chance to meet their pairs (default is 200000 reads); increasing the size reduces the number of temporary files created

--io-buffer-size=BUFFERSIZE
controls sizes of two buffers of BUFFERSIZE megabytes each, used for reading and writing BAM during the second pass (default is 128)

测试

去重复特别快，3G的bam文件去重复时间只用了1min。