按行进行单因素方差分析

Question

I would like to do an ANOVA analysis to compare the differences between the 4 groups (ABCD). Each group contains an uneven number of replicates in different columns, and each row represents one individual item I would like to test. Each column you can treat as a replicate and there is no relationship between different rows. Eventually, I hope to figure out what item(row) showed significant differences between the 4 groups.

Please see below example data structure. In reality, all data are normalized already.

	A1	A2	A3	B1	B2	C1	C2	D1	D2	D3
protein1	15	30	28	6	7	9	30	45	66	43
protein2	2	4	3	56	54	23	25	12	13	5
protein3	2	4	3	56	54	23	25	12	13	5
protein4	2	4	3	56	54	23	25	12	13	5

Answer 1

One way to do this:

First reshape the data into a format that the model can handle. This uses the tidyverse package.

df_long <- df %>%
  pivot_longer(cols = 2:ncol(.)) %>%
  pivot_wider(names_from = prot, values_from = value) %>%
  separate(name, into = c("trt"), sep = "\\d")

Which looks like:

   trt   protein1 protein2 protein3 protein4
   <chr>    <dbl>    <dbl>    <dbl>    <dbl>
 1 A           15        2        2        2
 2 A           30        4        4        4
 3 A           28        3        3        3
 4 B            6       56       56       56
 5 B            7       54       54       54
 6 C            9       23       23       23
 7 C           30       25       25       25
 8 D           45       12       12       12
 9 D           66       13       13       13
10 D           43        5        5        5

Then you can easily use whatever model/statistical test you would like to apply. For example, to generate an ANOVA for each column, you could define a helper function and then map over the columns:

fit_aov <- function(col) {
  aov(col ~ trt, data = df_long)
}

anovas <- map(df_long[, 2:ncol(df_long)], fit_aov)

summary(anovas$protein2)

            Df Sum Sq Mean Sq F value   Pr(>F)    
trt          3   3648  1216.0   165.8 3.69e-06 ***
Residuals    6     44     7.3                     
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1