将成对距离表转换为仅两列中个人的距离列表
Posted
技术标签:
【中文标题】将成对距离表转换为仅两列中个人的距离列表【英文标题】：Convert pairwise distance table to list of distance for individuals in only two columns 【发布时间】：2019-12-01 12:09:45 【问题描述】：我想将成对距离表（2 列中的观察值）转换为列出个人的表（1 列中的观察值）。基本上成对关系的信息会丢失（这与我的分析无关），并且它们各自的行的距离值需要加倍。
我可以用这段代码分隔字符串：
pairwise_readout <- str_split_fixed(pairwise[,1], " ", 4) #splits strings apart
pairwise_readout <- data.frame(pairwise_readout,pairwise$dist) #places distance again
但是不知道如何继续将表格重新排列成更少的列。所有搜索结果都只显示与成对表相关的解决方案。
这是一个示例数据集：
需要注意的重要一点是，我对每个观察的字符串中包含的“gr#”也很感兴趣。
pairwise <- data.frame(ind_comp = c("OP2645ii_d gr3 OP5048___g gr2","OP5046___e gr5 OP5048___g gr2","OP2413iiia gr1 OP5048___g gr2","OP5043___b gr1 OP5048___g gr2", "OP3088i___a gr1 OP5048___g gr2","OP5046___a gr5 OP5048___g gr2", "OP5048___b gr5 OP5048___g gr2", "OP5043___a gr3 OP5048___g gr2", "OP2645ii_d gr3 OP5048___g gr2", "OP2645ii_d gr3 OP5044___c gr2", "OP2413iiib gr4 OP5048___g gr2", "OP5046___c gr1 OP5048___g gr2"), dist = c(7.590363,6.449676,6.419955,6.349918,6.182623,6.162655,6.154232,6.140147,6.058633,5.962923,5.943956,5.863753))
基本上我想要一个遵循这种形式的表格：
pairwise_table_less_columns <- data.frame(ind_comp = c("OP2645ii_d","OP5048___g","OP5046___e", "OP5048___g", "OP2413iiia", "OP5048___g", "OP5043___b", "OP5048___g", "OP3088i___a", "OP5048___g", "OP5046___a", "OP5048___g", "OP5048___b", "OP5048___g", "OP5043___a", "OP5048___g", "OP2645ii_d", "OP5048___g", "OP2645ii_d", "OP5044___c", "OP2413iiib", "OP5048___g", "OP5046___c", "OP5048___g"), gr = c("gr3","gr2","gr5", "gr2", "gr1", "gr2", "gr1", "gr2", "gr1", "gr2", "gr5", "gr2", "gr5", "gr2", "gr3", "gr2", "gr3", "gr2", "gr3", "gr2", "gr4", "gr2", "gr1", "gr2"), dist = c(7.590363,7.590363,6.449676,6.449676,6.419955,6.419955,6.349918,6.349918,6.182623,6.182623,6.162655,6.162655,6.154232,6.154232,6.140147,6.140147,6.058633,6.058633,5.962923,5.962923,5.943956,5.943956,5.863753,5.863753))
【问题讨论】：
【参考方案1】：我们可以使用dplyr 和tidyr。首先separateind_comp基于空格分成4个不同的列，gather它变成长格式，从key列中删除数字，使它们具有相同的名称，使用row_number()创建一个公共标识符，然后spread到宽格式。
library(dplyr)
library(tidyr)
pairwise %>%
separate(ind_comp, c("ind_comp1", "gr1", "ind_comp2", "gr2"), sep = "\\s+") %>%
gather(key, value, -dist) %>%
mutate(key = sub("\\d+", "", key)) %>%
group_by(key) %>%
mutate(row = row_number()) %>%
spread(key, value) %>%
dplyr::select(-row)
# A tibble: 24 x 3
# dist gr ind_comp
# <dbl> <chr> <chr>
# 1 5.86 gr1 OP5046___c
# 2 5.86 gr2 OP5048___g
# 3 5.94 gr4 OP2413iiib
# 4 5.94 gr2 OP5048___g
# 5 5.96 gr3 OP2645ii_d
# 6 5.96 gr2 OP5044___c
# 7 6.06 gr3 OP2645ii_d
# 8 6.06 gr2 OP5048___g
# 9 6.14 gr3 OP5043___a
#10 6.14 gr2 OP5048___g
# … with 14 more rows
【讨论】：
【参考方案2】：另一个想法是用另一个分隔符替换第二个空格，然后拆分，即
library(dplyr)
library(tidyr)
pairwise %>%
mutate(ind_comp = gsub('([^ ]+ [^ ]+) ', '\\1|', ind_comp)) %>%
separate_rows(ind_comp, sep = '[|]')
给出，
ind_comp dist 1 OP2645ii_d gr3 7.590363 2 OP5048___g gr2 7.590363 3 OP5046___e gr5 6.449676 4 OP5048___g gr2 6.449676 5 OP2413iiia gr1 6.419955 6 OP5048___g gr2 6.419955 7 OP5043___b gr1 6.349918 8 OP5048___g gr2 6.349918 9 OP3088i___a gr1 6.182623 10 OP5048___g gr2 6.182623 11 OP5046___a gr5 6.162655 12 OP5048___g gr2 6.162655 13 OP5048___b gr5 6.154232 14 OP5048___g gr2 6.154232 15 OP5043___a gr3 6.140147 16 OP5048___g gr2 6.140147 17 OP2645ii_d gr3 6.058633 18 OP5048___g gr2 6.058633 19 OP2645ii_d gr3 5.962923 20 OP5044___c gr2 5.962923 21 OP2413iiib gr4 5.943956 22 OP5048___g gr2 5.943956 23 OP5046___c gr1 5.863753 24 OP5048___g gr2 5.863753
【讨论】：
【参考方案3】：这是一个基本的 R 解决方案。
将数据框pairwise_readout 分成两列然后rbind 它们。有一些中间步骤可以确保列名相同并对结果进行排序。
tmp1 <- pairwise_readout[c(1, 2, 5)]
tmp2 <- pairwise_readout[c(3, 4, 5)]
names(tmp1) <- names(tmp2) <- c("ind_comp", "gr", "dist")
tmp1$id <- tmp2$id <- seq_len(nrow(tmp1))
tmp <- rbind(tmp1,tmp2)
result <- tmp[order(tmp$id), -4]
最后清理。
rm(tmp, tmp1, tmp2)
【讨论】：
【参考方案4】：我迟到了，但这是我的解决方案：
library("stringr") #For str_split
pairwise <- data.frame(ind_comp = c("OP2645ii_d gr3 OP5048___g gr2","OP5046___e gr5 OP5048___g gr2","OP2413iiia gr1 OP5048___g gr2","OP5043___b gr1 OP5048___g gr2", "OP3088i___a gr1 OP5048___g gr2","OP5046___a gr5 OP5048___g gr2", "OP5048___b gr5 OP5048___g gr2", "OP5043___a gr3 OP5048___g gr2", "OP2645ii_d gr3 OP5048___g gr2", "OP2645ii_d gr3 OP5044___c gr2", "OP2413iiib gr4 OP5048___g gr2", "OP5046___c gr1 OP5048___g gr2"), dist = c(7.590363,6.449676,6.419955,6.349918,6.182623,6.162655,6.154232,6.140147,6.058633,5.962923,5.943956,5.863753))
pairwise$ind_comp <- as.character(pairwise$ind_comp)
pairwise$ind_comp2 <- sapply(str_split(pairwise$ind_comp, "(?<=\\s[a-z]2[0-9]1)\\s"), "[", 2) #Splitting to create second column
pairwise$ind_comp <- sapply(str_split(pairwise$ind_comp, "(?<=\\s[a-z]2[0-9]1)\\s"), "[", 1) #And first column
tmp_pairwise <- data.frame(ind_comp = pairwise$ind_comp2, dist = as.numeric(pairwise$dist)) #Copying second columna and corresponding distances to temporary object
pairwise <- pairwise[, -3] #Removing second column from original data frame
pairwise <- rbind(pairwise, tmp_pairwise) #Binding original data frame and the temporary data frame by rows
rm(tmp_pairwise) #Removing temporary data frame
pairwise$gr <- sapply(str_split(pairwise$ind_comp, "(?<=\\s)"), "[", 2) #Creating group column
pairwise$ind_comp <- sapply(str_split(pairwise$ind_comp, "(?<=\\s)"), "[", 1) #Fixing first column to remove group information
head(pairwise)
ind_comp dist gr
1 OP2645ii_d 7.590363 gr3
2 OP5046___e 6.449676 gr5
3 OP2413iiia 6.419955 gr1
4 OP5043___b 6.349918 gr1
5 OP3088i___a 6.182623 gr1
6 OP5046___a 6.162655 gr5
【讨论】：
以上是关于将成对距离表转换为仅两列中个人的距离列表的主要内容，如果未能解决你的问题，请参考以下文章
如何在没有交叉产品的情况下从 BigQuery 中的两列中取消嵌套两个列表，作为单独的行