Turning multiple rows into a string by merging on ID in R - r

Question

Table1:(there are hundreds of IDs)
participant_id hpo_term year_of_birth affected_relative genome
123 kidney failure 2000 Y 38
123 hand tremor 2000 Y 38
123 kidney transplant 2000 Y 38
432 hypertension 1980 N 37
432 exotropia 1980 N 37
432 scissor gait 1980 N 37
I have two look-up tables:(with hundreds of values in each)
Renal lookup:
kidney failure
kidney transplant
hypertension
Non-renal lookup(with hundreds of values in each):
hand tremor
exotropia
scissor gait
Desired outcome:
participant_id kidney_hpo_term non_kidney_hpo_term year_of_birth affected_relative genome
123 kidney failure;kidney transplant hand tremor 2000 Y 38
432 hypertension exotropia;scissor gait 1980 Y 37
Initially I tried:
library(dplyr); library(tidyr)
pt.data %>%
mutate(kidney = hpo_term %in% kidney.hpo) %>%
pivot_wider(names_from = kidney, values_from = hpo_term,
values_fn = function(x)paste(x,collapse = ";"), values_fill = NA) %>%
setNames(c("participant_id","Kidney","Non.kidney"))
with kidney.hpo <- read.delim("kidney_hpo_terms.txt", header = F)
But I get "Error in values_fn[[value]] ; object of type 'closure' is not subsettable"
Not sure what I am doing wrong and your help would be much appreciated.

Answer 1

There are several things to say with your data.
First, your table1 has duplicated columns: year_of_birth, affected_relative, and genome are the same for a given participant.
This should better be stored in a separate table, which I named table1_short.
For your very question, it is only a matter of checking whether a term is in a vector, which is done using %in%.
Here is how you could write the code:
library(tidyverse)
table1=read.table(header=T, text="
participant_id hpo_term year_of_birth affected_relative genome
123 'kidney failure' 2000 Y 38
123 'hand tremor' 2000 Y 38
123 'kidney transplant' 2000 Y 38
432 hypertension 1980 N 37
432 exotropia 1980 N 37
432 'scissor gait' 1980 N 37")
table1_short = table1 %>% select(-hpo_term) %>% group_by(participant_id) %>% slice(1)
table1_long = table1 %>% select(1:2)
renal_lookup = c("kidney failure", "kidney transplant", "hypertension")
nonrenal_lookup = c("hand tremor", "exotropia", "scissor gait")
table1_long %>%
group_by(participant_id) %>%
summarise(
kidney_hpo_term = hpo_term[hpo_term %in% renal_lookup] %>% paste(collapse=";"),
non_kidney_hpo_term = hpo_term[hpo_term %in% nonrenal_lookup] %>% paste(collapse=";")
) %>%
left_join(table1_short, by="participant_id")
#> # A tibble: 2 x 6
#> participant_id kidney_hpo_term non_kidney_hpo_term year_of_birth affected_relative genome
#> <int> <chr> <chr> <int> <chr> <int>
#> 1 123 kidney failure;kidney transplant hand tremor 2000 Y 38
#> 2 432 hypertension exotropia;scissor gait 1980 N 37
Created on 2021-05-12 by the reprex package (v2.0.0)

Answer 2

This can be done with dcast in data.table as follows:
dtt[, group := paste0(
ifelse(hpo_term %in% kidney_hpo, 'kidney', 'non_kidney'), '_hpo_term')]
dcast(dtt, ... ~ group, value.var = 'hpo_term',
fun.aggregate = paste, collapse = ';')
# participant_id year_of_birth affected_relative genome kidney_hpo_term
# 1: 123 2000 Y 38 kidney failure;kidney transplant
# 2: 432 1980 N 37 hypertension
# non_kidney_hpo_term
# 1: hand tremor
# 2: exotropia;scissor gait"