如何在R中合并两个带有“洞”的表

Question

我有来自三种不同实验条件、两种不同模式的数据，我已经计算并存储在第一个 data.frame 中，每个模式的每个条件的平均值和标准差。好像

t1 <- data.frame(condition = rep(c("A","B","C"),times=2),
             mode = rep(c(1,2),each=3),
             mean = rnorm(6),
             sd = rnorm(6))

然后，我计算了 mann-whitney 秩和检验的每个成对比较条件。生成的 data.frame 如下所示：

t2 <- data.frame(mode = rep(c(1,2),each=3),
             test = rep(c("AvB","AvC","BvC"),2),
             pvalue = runif(6,0,0.2)
             )

我想将两个 data.frames 与“holes”组合在一个 data.frame 中，每个测试显示模式、实际测试条件的均值和 p 值。 我想到了这样的事情：

mode    mean A  mean B  mean C  test    pvalue
1       1.34    1.12            A v B   0.067
1       1.34            0.98    A v C   0.021
1               1.12    0.98    B v C   0.345

我知道结果表中存在漏洞——不太适合在其上运行统计信息——但我的目标是创建一个 .tex 表（使用 xtable()）来显示所有可能的成对测试，并且，同时显示每个模式下每个条件下变量的均值。这是为了在纸质附录中展示。

谢谢！

Answer 1

没有使用 reshape2 的 plyr 或 dplyr 的解决方案。它可以处理更多的条件和测试，并且不需要硬编码比较。

library(reshape2)
#make column with variables that are tested (so can be used if many pairs/changing
#names as long as there is a clear separation
t2$var1 <- gsub("v.+","",t2$test)
t2$var2 <- gsub(".+v","",t2$test)

#combine data in long format
t3 <- merge(t1,t2,by="mode")

#create mean_na vector with removed means when test is not performed
t3$mean_na <- with(t3, ifelse(condition==var1|condition==var2,mean,NA))

#cast to table
res_temp <- dcast(mode+test~condition,value.var="mean_na",data=t3)
res_temp

#re-add pvalue
result_final <- merge(res_temp,t2[,c("mode","test","pvalue")],by=c("mode","test"))

> result_final
  mode test           A          B        C     pvalue
1    1  AvB -0.56047565 -0.2301775       NA 0.13114116
2    1  AvC -0.56047565         NA 1.558708 0.14170609
3    1  BvC          NA -0.2301775 1.558708 0.10881320
4    2  AvB  0.07050839  0.1292877       NA 0.11882840
5    2  AvC  0.07050839         NA 1.715065 0.05783195
6    2  BvC          NA  0.1292877 1.715065 0.02942273

Answer 2

这是一个想法：

library(tidyr)
library(dplyr)

t2 %>% 
  separate(test, into = c("cond1", "cond2"), sep = "v", remove = FALSE) %>%
  gather(key, condition, -mode, -pvalue, -test) %>%
  select(-key) %>%
  left_join(t1 %>% select(-sd)) %>% 
  spread(condition, mean) %>%
  setNames(c(names(.)[1:3], paste("mean", names(.)[4:6])))

这给出了：

#  mode test      pvalue     mean A    mean B     mean C
#1    1  AvB 0.053444134 -0.6264538 0.1836433         NA
#2    1  AvC 0.077222819 -0.6264538        NA -0.8356286
#3    1  BvC 0.002678067         NA 0.1836433 -0.8356286
#4    2  AvB 0.076477591  1.5952808 0.3295078         NA
#5    2  AvC 0.173938169  1.5952808        NA -0.8204684
#6    2  BvC 0.068069799         NA 0.3295078 -0.8204684

如果您想将结果传递到 xtable 中，您可以添加：

... %>% xtable() %>% print(type = "html")

<!-- html table generated in R 3.2.1 by xtable 1.7-4 package -->
<!-- Fri Jun 19 18:45:36 2015 -->
<table border=1>
<tr> <th>  </th> <th> mode </th> <th> test </th> <th> pvalue </th> <th> mean A </th> <th> mean B </th> <th> mean C </th>  </tr>
<tr> <td align="right"> 1 </td> <td align="right"> 1.00 </td> <td> AvB </td> <td align="right"> 0.05 </td> <td align="right"> -0.63 </td> <td align="right"> 0.18 </td> <td align="right">  </td> </tr>
<tr> <td align="right"> 2 </td> <td align="right"> 1.00 </td> <td> AvC </td> <td align="right"> 0.08 </td> <td align="right"> -0.63 </td> <td align="right">  </td> <td align="right"> -0.84 </td> </tr>
<tr> <td align="right"> 3 </td> <td align="right"> 1.00 </td> <td> BvC </td> <td align="right"> 0.00 </td> <td align="right">  </td> <td align="right"> 0.18 </td> <td align="right"> -0.84 </td> </tr>
<tr> <td align="right"> 4 </td> <td align="right"> 2.00 </td> <td> AvB </td> <td align="right"> 0.08 </td> <td align="right"> 1.60 </td> <td align="right"> 0.33 </td> <td align="right">  </td> </tr>
<tr> <td align="right"> 5 </td> <td align="right"> 2.00 </td> <td> AvC </td> <td align="right"> 0.17 </td> <td align="right"> 1.60 </td> <td align="right">  </td> <td align="right"> -0.82 </td> </tr>
<tr> <td align="right"> 6 </td> <td align="right"> 2.00 </td> <td> BvC </td> <td align="right"> 0.07 </td> <td align="right">  </td> <td align="right"> 0.33 </td> <td align="right"> -0.82 </td> </tr>
</table>

---

数据

set.seed(1)

t1 <- data.frame(condition = rep(c("A","B","C"),times=2),
                 mode = rep(c(1,2),each=3),
                 mean = rnorm(6),
                 sd = rnorm(6))

t2 <- data.frame(mode = rep(c(1,2),each=3),
                 test = rep(c("AvB","AvC","BvC"),2),
                 pvalue = runif(6,0,0.2))

Answer 3

这适用于相当小的数据；如果你真的有更多的条件，这种方法不会很有趣：

library(reshape2)
library(dplyr)

means <- dcast(t1, mode~condition, value.var='mean')

results <- merge(t2, means) %>%
    # extract conditions of each test row
    mutate(first=substr(test,1,1), second=substr(test,3,3), 
           # replace untested condition means with ""
              meanA=ifelse(first=='A'|second=='A', A, ""),
              meanB=ifelse(first=='B'|second=='B', B, ""),
              meanC=ifelse(first=='C'|second=='C', C, "")) %>%
    # discard unwanted columns
    select(mode, meanA, meanB, meanC, test, pvalue)

> print(results)
  mode  meanA  meanB meanC test     pvalue
1    1 -0.646 -0.778        AvB 0.03597074
2    1 -0.646        0.014  AvC 0.11394383
3    1        -0.778 0.014  BvC 0.14393726
4    2  0.083  0.275        AvB 0.16493966
5    2  0.083        0.619  AvC 0.06548713
6    2         0.275 0.619  BvC 0.08474118