将数据转换为正确的INEXT分析格式

Question

我计划使用iNEXT功能（来自iNEXT包）来分析两种池塘类型之间的潜在差异。这些数据是存在/不存在和按物种矩阵的位点，例如

Pond.type <- c("A", "A", "A", "B", "B", "B")
Sample.no <- c(1,2,3,1,2,3)
Species1 <- c(0,1,1,1,0,0)
Species2 <- c(0,1,1,0,1,0)
Species3 <- c(1,1,1,1,0,1)
Species4 <- c(0,1,0,1,0,0)
Species5 <- c(1,1,1,0,0,0)

mydata <- cbind.data.frame(Pond.type, Sample.no, Species1, Species2, Species3, Species4, Species5)

如果我将数据框分成池类型并将它们保存为矩阵，我可以运行该函数，例如iNEXT(pond.type.a.dataframe)，但我无法在一个地块上比较不同的池塘类型。

我的问题是，有没有办法将我的数据转换为与iNEXT库中提供的ciliates示例数据相同的格式？这是作为矩阵列表给出的。

Answer 1

如果您有2个像这样结构的.csvs

A型池塘：

Species, Sample1, Sample2, Sample3
Species1, 0,1,1
Species2, 0,1,1
Species3, 1,1,1
Species4, 0,1,0
Species5, 1,1,1

B型池塘：

Species, Sample1, Sample2, Sample3
Species1, 1,0,0
Species2, 0,1,0
Species3, 1,0,1
Species4, 1,0,0
Species5, 0,0,0

然后假设您调用“读取csv”的池塘A和池塘B，您可以执行以下操作：

library(iNEXT)
library(ggplot2)

# make a matrix from pondA as type "integer"
mPondA <- as.matrix(apply(pondA[,-1],2,as.integer))

# use your species names as row names
row.names(mPondA) <- pondA[,1]

# do the same for pondB
mPondB <- as.matrix(apply(pondB[,-1],2,as.integer))
row.names(mPondB) <- pondB[,1]

# create a list of your matrices (named so the output looks nice)
pondABM = list(A = mPondA, B = mPondB)

# have a look at the raw data
out.raw <- iNEXT(pondABM, datatype="incidence_raw", endpoint=20)
ggiNEXT(out.raw)

这将为您提供如下输出：

如果您想使用此输入类型对数据执行更多操作，请不要忘记更改数据类型：

# note that datatype has been changed to "incidence_raw" instead of "abundance"
iNEXT(pondABM, q=0, datatype="incidence_raw", size=NULL, se=TRUE, conf=0.95, nboot=50)

Answer 2

我尝试了Oliver Burdekin的答案，效果很好 - 谢谢

使用丰度数据时要做的一个补充 - 在运行iNEXT函数之前将 as.abucount 函数应用于矩阵列表

[1] 08 02 2017 13:03:20.768:INFO [karma]: Karma v1.4.1 server started
... 
[1] 08 02 2017 13:03:22.579:WARN [web-server]: 404: /app.component.html
    Chrome 56.0.2924 (Linux 0.0.0) AppComponent should create component FAILED
[1]     Error: This test module uses the component AppComponent which is using a "templateUrl" or "styleUrls", but they were never compiled. Please call "TestBed.compileComponents" before your test.
[1]     Expected undefined to be defined.

Answer 3

我找到了一个解决方案，可能比它需要的更啰嗦，但它确实有效。

# I've altered the dataset as I usually type these data in long format
Pond.type <- c(rep("A", 15), rep("B", 15)) 
Site.no <- rep(seq(1,6, by=1), 5)
Species <- c(rep("Spp1", 6), rep("Spp2", 6), rep("Spp3", 6), rep("Spp4", 6), rep("Spp5", 6))
Presence <- rep(1, 30)

# join the vectors together into a dataframe
mydata.long <- cbind.data.frame(Pond.type, Site.no, Species, Presence)

# I then cast the data into a species x site matrix (dcast is from the reshape2 library)
mydata.cast <- dcast(mydata.long, Species + Pond.type ~ Site.no)

# Changes the NAs to zeros. 
mydata.cast[is.na(mydata.cast)] <- 0

# Separate the dataframe into each pond tyoe
pondA <- mydata.cast[grep("A", mydata.cast$Pond.type),]
pondB <- mydata.cast[grep("B", mydata.cast$Pond.type),]

# Calculate the frequency counts using the  function from the iNEXT library
pondA.freq <- as.incfreq(pondA[,3:ncol(pondA)])
pondB.freq <- as.incfreq(pondB[,3:ncol(pondB)])

# join them as a list
pond.list = list(A = pondA.freq, B = pondB.freq)

# ready for comparison
pond.out <- iNEXT(pond.list, datatype = "incidence_freq", q=0)
pond.out