具有分类变量的多个表的欧几里得距离计算

Question

我有两个如下所示的数据框：

df1 <- data.frame(geneID=c("gene1","gene2","gene3","gene4",
                           "gene5","gene6","gene7","gene8","gene9","gene10"),
                  patient_ID=c(700,0,3,387,30724,1,609,4,0,1729))
    
head(df1)
   geneID patient_ID
1  gene1        700
2  gene2          0
3  gene3          3
4  gene4        387
5  gene5      30724
6  gene6          1   

df2 <- data.frame(component1=c("gene1","gene2","gene3","gene4","gene5"),
                  component2=c("gene2","gene4","gene5","gene10","gene9"))
head(df2)
    component1 component2
1      gene1      gene2
2      gene2      gene4
3      gene3      gene5
4      gene4     gene10
5      gene5      gene9   

我想生成一个使用来自 df1 的基因值的数据框，并从 df2 计算组件 1 和 2 之间的欧几里得距离。例如，对于gene3 和gene5 对，df3 中的输出应使用以下等式计算：

val = sqrt((gene3)^2+(gene5)^2) =sqrt(700^2+30724^2)

我的最终目标是得到这样的桌子：

       gene1 gene2 gene3 gene4 gene5 gene6 gene7 gene8 gene9 gene10
1   gene1     0     0     0     0     0     0     0     0     0      0
2   gene2   val     0     0     0     0     0     0     0     0      0
3   gene3     0     0     0     0     0     0     0     0     0      0
4   gene4     0   val     0     0     0     0     0     0     0    val
5   gene5     0     0   val     0     0     0     0     0   val      0
6   gene6     0     0     0     0     0     0     0     0     0      0
7   gene7     0     0     0     0     0     0     0     0     0      0
8   gene8     0     0     0     0     0     0     0     0     0      0
9   gene9     0     0     0     0   val     0     0     0     0      0
10 gene10     0     0     0   val     0     0     0     0     0      0

我非常感谢任何帮助和建议。

谢谢！

欧哈

Answer 1

试试这个。

library(dplyr)
library(tidyr) # pivot_wider
left_join(df2, select(df1, geneID, x = patient_ID), by = c("component1" = "geneID")) %>%
  left_join(select(df1, geneID, y = patient_ID), by = c("component2" = "geneID")) %>%
  mutate(val = sqrt(x^2 + y^2)) %>%
  complete(component1, component2) %>%
  pivot_wider(component1, names_from = component2, values_from = val)
# # A tibble: 5 x 6
#   component1 gene10 gene2 gene4  gene5 gene9
#   <chr>       <dbl> <dbl> <dbl>  <dbl> <dbl>
# 1 gene1         NA    700    NA    NA     NA
# 2 gene2         NA     NA   387    NA     NA
# 3 gene3         NA     NA    NA 30724.    NA
# 4 gene4       1772.    NA    NA    NA     NA
# 5 gene5         NA     NA    NA    NA  30724

我没有尝试将其扩展为在两个轴上都有 1-10，因为您的 df2 建议特定配对。您可能可以使用tidyr::complete 或tidyr::expand 来获得完整的扩展。