r - 如何使用 R 在 t.test 中仅运行比较的子集？

Question

我正在使用 R 来做一些统计，这个问题是从统计交换中重复的，它被关闭，因为它不是真正的统计问题，所以我认为它可能与堆栈溢出( https://stats.stackexchange.com/questions/441638/how-do-i-run-only-a-subset-of-comparisons-in-a-t-test-using-r/441674#441674 )更相关。尽管这里给出的答案(对数据进行子集化，然后运行测试)在逻辑上似乎是正确的，但我看不出有什么方法可以在不对每个聚糖重复 100 段不同的代码的情况下做到这一点(见下文):

我已经从原始数据生成了一个data.frame。数据包括一个数值变量 (fold_change) 和两个因子变量(dis_status 包括 RF 和 con，以及聚糖，其中包括 100 种不同的聚糖)

这是一个可重现的示例，其中只有 3 个聚糖，每个聚糖有 3 个“RF”和 3 个“con”。

   > example
   dis_status glycan fold_change
1          RF      a  4.83433185
2          RF      a  3.88519084
3          RF      a  2.80368849
4         con      a  0.94730194
5         con      a  1.91278688
6         con      a  1.23225002
7          RF      b  4.07173876
8          RF      b  5.70383491
9          RF      b  0.05282291
10        con      b  1.34631723
11        con      b  4.26723583
12        con      b  4.26723583
13         RF      c  2.20887813
14         RF      c  4.62220094
15         RF      c  0.94730194
16        con      c  0.53597973
17        con      c  2.92572685
18        con      c  1.58871049

> dput(example)
structure(list(dis_status = structure(c(2L, 2L, 2L, 1L, 1L, 1L, 
2L, 2L, 2L, 1L, 1L, 1L, 2L, 2L, 2L, 1L, 1L, 1L), .Label = c("con", 
"RF"), class = "factor"), glycan = structure(c(1L, 1L, 1L, 1L, 
1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, 3L, 3L, 3L), .Label = c("a", 
"b", "c"), class = "factor"), fold_change = c(4.834331853, 3.885190842, 
2.803688487, 0.947301944, 1.912786879, 1.232250023, 4.071738761, 
5.703834911, 0.052822912, 1.346317234, 4.267235834, 4.267235834, 
2.208878135, 4.622200944, 0.947301944, 0.535979733, 2.925726849, 
1.588710491)), class = "data.frame", row.names = c(NA, -18L))

我可以对数据运行 t.test:

ad_nonpaired <- pairwise.t.test(stats_df$fold_change,  stats_df$dis_status:stats_df$glycan, 
                               paired = F,
                               pool.sd = F,
                               p.adj = "none")

接下来我将纠正多重比较，但我遇到的问题是在 dis_status 和聚糖的每种可能组合之间进行 t.tests。

我只对每个聚糖的“RF”与“con”感兴趣。因此，对于上面的三个聚糖，我只想将“RF”中的“x”与“con”中的“x”进行比较，而不是“x”与“y”之间的任何比较，但无法弄清楚如何在测试中指定这一点？

R version 3.5.2 (2018-12-20)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: macOS Mojave 10.14.6

Matrix products: default
BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/3.5/Resources/lib/libRlapack.dylib

locale:
[1] en_NZ.UTF-8/en_NZ.UTF-8/en_NZ.UTF-8/C/en_NZ.UTF-8/en_NZ.UTF-8

attached base packages:
[1] parallel  stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] knitr_1.25          broom_0.5.2         ggrepel_0.8.1       readxl_1.3.1        forcats_0.4.0       stringr_1.4.0       dplyr_0.8.3         purrr_0.3.3        
 [9] readr_1.3.1         tidyr_1.0.0         tibble_2.1.3        ggplot2_3.2.1       tidyverse_1.2.1     limma_3.38.3        hexbin_1.27.3       vsn_3.50.0         
[17] Biobase_2.42.0      BiocGenerics_0.28.0

loaded via a namespace (and not attached):
 [1] Rcpp_1.0.2            lubridate_1.7.4       lattice_0.20-38       gtools_3.8.1          rprojroot_1.3-2       assertthat_0.2.1      zeallot_0.1.0         digest_0.6.22        
 [9] utf8_1.1.4            plyr_1.8.4            R6_2.4.0              cellranger_1.1.0      backports_1.1.5       evaluate_0.14         highr_0.8             httr_1.4.1           
[17] pillar_1.4.2          gplots_3.0.1.1        zlibbioc_1.28.0       rlang_0.4.1           lazyeval_0.2.2        curl_4.2              rstudioapi_0.10       gdata_2.18.0         
[25] preprocessCore_1.44.0 desc_1.2.0            labeling_0.3          splines_3.5.2         munsell_0.5.0         xfun_0.10             compiler_3.5.2        modelr_0.1.5         
[33] pkgconfig_2.0.3       tidyselect_0.2.5      fansi_0.4.0           crayon_1.3.4          withr_2.1.2           bitops_1.0-6          grid_3.5.2            nlme_3.1-141         
[41] jsonlite_1.6          gtable_0.3.0          lifecycle_0.1.0       affy_1.60.0           magrittr_1.5          scales_1.0.0          KernSmooth_2.23-16    cli_1.1.0            
[49] stringi_1.4.3         affyio_1.52.0         testthat_2.2.1        xml2_1.2.2            ellipsis_0.3.0        generics_0.0.2        vctrs_0.2.0           tools_3.5.2          
[57] glue_1.3.1            hms_0.5.2             pkgload_1.0.2         yaml_2.2.0            colorspace_1.4-1      BiocManager_1.30.9    caTools_1.17.1.2      rvest_0.3.4          
[65] haven_2.1.1          

Answer 1

您可以按聚糖拆分数据框，然后按 dis_status 组进行 t 测试，无需任何外部库:

results <- do.call("rbind", lapply(split.data.frame(df, df$glycan), 
                  function(x) {
                    pairwise.t.test(x$fold_change, x$dis_status,
                                    paired = FALSE, pool.sd = FALSE, 
                                    p.adj = "none") -> test;
                    as.numeric(tapply(x$fold_change, x$dis_status, mean)) -> ta;
                    data.frame(glycan = as.character(x$glycan[1]), 
                               mean.con = ta[1],
                               mean.RF = ta[2],
                               pvalue = as.numeric(test$p.value));
                  }))

根据评论给出您想要的数据框

results
  glycan mean.con  mean.RF     pvalue
a      a 1.364113 3.841070 0.03403083
b      b 3.293596 3.276132 0.99335164
c      c 1.683472 2.592794 0.52325471