r - 结合 OTU 和税表并用 OTU id 替换实际序列(Phyloseq/dada2)

Question

我正在遵循此处描述的工作流程 https://f1000research.com/articles/5-1492/v2使用示例数据以及我自己的数据。这工作正常，但现在我无法生成 OTU 表，其中包含诸如“OTU00004”或更好的“kingdom_phylum_..._Pseudomonas_OTU00004”之类的标题。我想使用这样的表格来查找和绘制多个样本中某个 OTU 的丰度。

我创建了一个名为 ps 的对象，这似乎没问题:

ps <- phyloseq(tax_table(taxtab), sample_data(samdf),
                 otu_table(seqtab, taxa_are_rows = FALSE),phy_tree(fitGTR$tree))    

> ps
    phyloseq-class experiment-level object
    otu_table()   OTU Table:         [ 454 taxa and 360 samples ]
    sample_data() Sample Data:       [ 360 samples by 14 sample variables ]
    tax_table()   Taxonomy Table:    [ 454 taxa by 6 taxonomic ranks ]
    phy_tree()    Phylogenetic Tree: [ 454 tips and 452 internal nodes ]

但是 OTU 表中的标题和分类表中的相应行是实际的(这里是缩短的)序列

> head(otu_table(ps)[1])
     GCAAGCGTTACTCGGAATCACTGGGCGTAAAGAGCGCGTAGGCGG#shortened
F3D0                                             0

> head(tax_table(ps)[1])
Taxonomy Table:     [1 taxa by 6 taxonomic ranks]:
                                                         Kingdom
GCGAGCGTTATCCGGATTTATTGGGTTTAAAGGGTGCGCAGGCGGA#shortened "Bacteria"

有没有办法结合 otu 表和分类表中的信息并用编号的 OTU id 替换序列？我检查了几个 phyloseq 资源和常见问题解答，但我找不到答案。

我想要一张看起来像这样的表:

        taxonomy_OTU00001   taxonomy_OTU00002   taxonomy_OTU00003
F3D0    #counts             #counts             #counts
F3D1    #counts             #counts             #counts
F3D11   #counts             #counts             #counts
F3D125  #counts             #counts             #counts

由于此步骤之前的工作流程非常耗时，我不确定如何为此问题提供可重现的示例。

编辑:我按照 dww 的建议生成了一个示例子集。

short_otu2 = short_otu = head(otu_table(ps)[,c(1:6)])  # seq as colnames 
short_tax2 = short_tax = tax_table(ps)[colnames(short_otu), ]  # seq as rownames
# shorten seqs, must still be unique
colnames(short_otu2) <- substr(colnames(short_otu), 0, 50)
rownames(short_tax2) <- substr(rownames(short_tax), 0, 50)

library(phyloseq)
> dput(short_otu2)
new("otu_table", .Data = structure(c(526L, 375L, 2931L, 994L,
2061L, 419L, 319L, 330L, 1737L, 623L, 1868L, 350L, 402L, 207L,
1880L, 577L, 887L, 303L, 413L, 64L, 838L, 698L, 939L, 484L, 146L,
126L, 496L, 440L, 1183L, 184L, 462L, 37L, 26L, 782L, 271L, 310L
), .Dim = c(6L, 6L), .Dimnames = list(c("F3D0", "F3D1", "F3D11",
"F3D125", "F3D13", "F3D141"), c("GCGAGCGTTATCCGGATTTATTGGGTTTAAAGGGTGCGCAGGCGGAAGAT",
"GCGAGCGTTATCCGGATTTATTGGGTTTAAAGGGTGCGCAGGCGGACTCT", "GCGAGCGTTATCCGGATTTATTGGGTTTAAAGGGTGCGTAGGCGGGCTGT",
"GCGAGCGTTATCCGGATTTATTGGGTTTAAAGGGTGCGTAGGCGGGCTTT", "CCGAGCGTTATCCGGATTTATTGGGTTTAAAGGGAGCGTAGGTGGATTGT",
"GCGAGCGTTATCCGGATTTATTGGGTTTAAAGGGTGCGTAGGCGGCCTGC"))), taxa_are_rows = FALSE)

> dput(short_tax2)
new("taxonomyTable", .Data = structure(c("Bacteria", "Bacteria",
"Bacteria", "Bacteria", "Bacteria", "Bacteria", "Bacteroidetes",
"Bacteroidetes", "Bacteroidetes", "Bacteroidetes", "Bacteroidetes",
"Bacteroidetes", "Bacteroidia", "Bacteroidia", "Bacteroidia",
"Bacteroidia", "Bacteroidia", "Bacteroidia", "Bacteroidales",
"Bacteroidales", "Bacteroidales", "Bacteroidales", "Bacteroidales",
"Bacteroidales", "Bacteroidales_S24-7_group", "Bacteroidales_S24-7_group",
"Bacteroidales_S24-7_group", "Bacteroidales_S24-7_group", "Bacteroidaceae",
"Bacteroidales_S24-7_group", NA, NA, NA, NA, "Bacteroides", NA
), .Dim = c(6L, 6L), .Dimnames = list(c("GCGAGCGTTATCCGGATTTATTGGGTTTAAAGGGTGCGCAGGCGGAAGAT",
"GCGAGCGTTATCCGGATTTATTGGGTTTAAAGGGTGCGCAGGCGGACTCT", "GCGAGCGTTATCCGGATTTATTGGGTTTAAAGGGTGCGTAGGCGGGCTGT",
"GCGAGCGTTATCCGGATTTATTGGGTTTAAAGGGTGCGTAGGCGGGCTTT", "CCGAGCGTTATCCGGATTTATTGGGTTTAAAGGGAGCGTAGGTGGATTGT",
"GCGAGCGTTATCCGGATTTATTGGGTTTAAAGGGTGCGTAGGCGGCCTGC"), c("Kingdom",
"Phylum", "Class", "Order", "Family", "Genus"))))

Answer 1

这是问题的一部分“用 OTU ids (Phyloseq/dada2) 替换实际序列？”

我联系了 phyloseq/dada2 开发人员，根据 Susan Holmes 的回复 (https://github.com/joey711/phyloseq/issues/1030)，我想出了这段代码，用编号的 OTU header 替换扩增子序列。

进一步的讨论可以在这里找到:https://github.com/joey711/phyloseq/issues/213

# this changes the header from the actual sequence to Seq_001, Seq_002 etc
taxa_names(ps)
n_seqs <- seq(ntaxa(ps))
len_n_seqs <- nchar(max(n_seqs))
taxa_names(ps) <- paste("Seq", formatC(n_seqs, 
                                            width = len_n_seqs, 
                                            flag = "0"), sep = "_")
taxa_names(ps)

将分类法​​包含在标题中的一种可能方法如下(从上面继续):

# generate a vector containing the full taxonomy path for all OTUs
wholetax <- do.call(paste, c(as.data.frame(tax_table(ps))
                  [c("Kingdom", "Phylum", "Class", "Order", "Family", "Genus")], 
                  sep = "__"))  # to distinguish from "_" within tax ranks

# turn the otu_table into a data.frame
otu_export <- as.data.frame(otu_table(ps))
tmp <- names(otu_export)

# paste wholetax and OTU_ids together
for(i in 1:length(tmp)){
names(tmp)[i] = paste(wholetax[i], tmp[i], sep = "__")
}

# overwrite old names
names(otu_export) <- names(tmp)

> head(otu_export)[5]

# output:  
     Bacteria__Bacteroidetes__Bacteroidia__Bacteroidales__Bacteroidaceae__Bacteroides__Seq_005
F3D0                                                                                         146
F3D1                                                                                         126
F3D11                                                                                        496
F3D125                                                                                       440
F3D13                                                                                       1183
F3D141                                                                                       184

这还不包括对表格之间正确排序的测试!因此，请确保粘贴和覆盖正确。

这样你就有了一个 data.frame，其中包含每个分类等级、OTU id、样本名称和一个文件中的计数的分类“可拆分”。但是除了导出文件之外，您仍然维护 phyloseq 结构，其中 OTU_id 链接不同的表，例如 otu_table() 和 tax_table()。另一种方法是提供 wholetax矢量到 taxa_names()命令，我没有测试过。

非常欢迎改进建议!