在数据子集的不同变量中查找最小值和最大值

Question

This is probably very simple for all of you guru programmers, but I'm new at R and I seek some help. First I'll try to describe my data and then ask a question. I have 110k obs. of 24 variables:

      qseqid    evalue    pident    lenght    .......
    1 LL_206    3e-22     65.7      612
    2 LL_206    5e-22     75.6      485
    3 LL_206    5e-14     80.6      598
    4 LL_300    4e-22     90.5      251
    5 LL_300    4e-22     64.7      589
    6 LL_300    8e-14     89.8      125
    .
    .
    .

Now you can see that my data has subsets at qseqid variable. What I'm trying to get from my data is to find min evalue, max pident and max lenght for each subset of qseqid variable.

My results should look like this:

      qseqid    evalue    pident    lenght    .......
    1 LL_206    3e-22     65.7      612
    2 LL_300    4e-22     90.5      251
    .
    .
    .

I want that results are presented as csv table and in table should also be included all variables. I tried aggregate method, but I don't know how to tell R to find first min evalue then max pident and so on. Your help would be much appreciated.

Answer 1

A solution using dplyr :

library(dplyr)
res <- group_by(your_data_frame, qseqid) %>%
  summarise(evalue = min(evalue), pident = max(pident), 
    length = max(length))

You can then save res using write.csv / write.csv2 .

Answer 2

The way I can think about it, is to create the function which make a subset of all rows that fulfill first requirement, then subset for the second and so on..

ComplexSubset <- function(id){
  data.t <- data[data$qseqid %in% id,] # subset the needed qseqid
  data.t <- data.t[data.t$eval %in% min(data.t$eval),]  # subset the ALL min eval
  data.t <- data.t[data.t$pident %in% max(data.t$pident),]  # subset ALL max pident
  data.t <- data.t[data.t$lenght %in% max(data.t$lenght),]  #subset ALL max lenght
#here you can add additional subseting for your remaining colums
  return(data.t) ## save the row ID
}

## And then run for all unique qseqid

data.final <- data[0,]

for(id in unique(data$qseqid)){
  data.final <- rbind(data.final, ComplexSubset(id))
}

This should work. Hope it helpful

Answer 3

Ok, this is not 100% what I want as output. But it's close...

qseqid <- c("LL_1", "LL_1", "LL_1", "LL_1", "LL_2", "LL_2", "LL_2", "LL_2")
evalue <- c(1e-34, 1e-34, 1e-34, 1e-25, 2e-85, 2e-85, 3e-85, 1e-80)
pident <- c(90.5, 90.5, 80.8, 90.5, 75.3, 85.6, 75.3, 65.2)
lenght <- c(485, 503, 897, 1052, 689, 4859, 50, 115)
title <- c("A", "B", "C", "D", "E", "F", "G", "H")
mojadata <- data.frame(qseqid, evalue, pident, lenght, title)

Data:

   qseqid evalue pident lenght title
1:   LL_1  1e-34   90.5    485     A
2:   LL_1  1e-34   90.5    503     B
3:   LL_1  1e-34   80.8    897     C
4:   LL_1  1e-25   90.5   1052     D
5:   LL_2  2e-85   75.3    689     E
6:   LL_2  2e-85   85.6   4859     F
7:   LL_2  3e-85   75.3     50     G
8:   LL_2  1e-80   65.2    115     H

Code:

mojadata = data.table (mojadata)
mojadataout<-mojadata[, list(evalue=min(evalue),
       pident=max(pident[evalue == min(evalue)]), 
       lenght=max(lenght[pident==max(pident)&evalue==min(evalue)])), 
       by=list(qseqid)]

Output (what I get):

   qseqid evalue pident lenght
1:   LL_1  1e-34   90.5    503
2:   LL_2  2e-85   85.6   4859 

Output (what I want):

   qseqid evalue pident lenght title
1:   LL_1  1e-34   90.5    503     B
2:   LL_2  2e-85   85.6   4859     F

Now, how do I put in output variable "title"? I hope this time my question is more understandable.