基于qiime2的扩增子数据分析
-
DADA2的限制

DADA2对于单体序列 无法识别 -
https://github.com/benjjneb/dada2/issues/1095
dereplication (derepFastq) required?
dada2在处理的时候 可以去重 只是为了降低内存消耗?在最后统计丰度的时候 会把重复再计算回来? -
dyn.load("./src/dada2.so")
Error in dyn.load("./src/dada2.so") :
unable to load shared object '/home/jynlix/Downloads/src/dada2/dada2/./src/dada2.so':
/home/jynlix/Downloads/src/dada2/dada2/./src/dada2.so: undefined symbol: _ZTIN3tbb4taskE使用gdb 调试dada2时报错 dada2依赖 Rcpp 做多线程 Rcpp 依赖 tbb 需要提前把这个包加载
dyn.load("/home/jynlix/R/x86_64-pc-linux-gnu-library/4.1/RcppParallel/lib/libtbb.so.2", local = FALSE)
-
https://btep.ccr.cancer.gov/docs/qiime2/index.html
Microbiome Analysis with QIIME2 教程 -
dada2的执行过程:
filterAndTrim 过滤,截取
derepFastq (可选步骤 dada内部会判断 如果没有经过去重 会自己调用这个函数)
learnErrors 这个函数也会调用dada 以 self-consist mode 模式运行
dada 核心步骤 但是 self-consist mode = FALSE -
https://forum.qiime2.org/t/difference-between-sub-sampling-and-rarefing/18219
sub-sampling and rarefing的对比:


sub-sampling可以从sample或者feature的角度 对数据进行过滤 得到的是整个数据的子集。
rarefy是对所有feature的结果进行归一化。 -
假如data2的代码保存在 /root/data2
cd /root
R CMD build dada2
会生成dada2.so -
-
https://www.jandonline.org/article/S2212-2672(20)30438-X/pdf
Associations between Diet, the Gut Microbiome,
and Short-Chain Fatty Acid Production among
Older Caribbean Latino Adults
饮食和微生物多样性的研究 主要用了关联性分析 -
https://drive5.com/usearch/manual/amplification_bias.html
Abundance and amplification bias in amplicon sequencing
序列的丰度和物种的实际丰度 可能偏差比较大