CODEWITHSUNDEEP
regexr
fra
fra

Reputation: 852

match a word an not the entire line in R

I have a text file formatted as such:

#    title "Secondary Structure"
#    xaxis  label "Time (ns)"
#    yaxis  label "Number of Residues"
#TYPE xy
# subtitle "Structure = A-Helix + B-Sheet + B-Bridge + Turn"
# view 0.15, 0.15, 0.75, 0.85
# legend on
# legend box on
# legend loctype view
# legend 0.78, 0.8
# legend length 2
# s0 legend "Structure"
# s1 legend "Coil"
# s2 legend "B-Sheet"
# s3 legend "B-Bridge"
# s4 legend "Bend"
# s5 legend "Turn"
# s6 legend "A-Helix"
# s7 legend "5-Helix"
# s8 legend "3-Helix"
# s9 legend "Chain_Separator"
   0   637   180   201     7    94   129   300     0    47     1
   1   617   189   191    11    99   121   294     5    48     1
   2   625   183   198     7    97   130   290     0    53     1
   3   625   180   195     5   102   125   300     0    51     1
   4   622   185   196     5    99   117   304     0    52     1
   5   615   192   190     5   106   121   299     0    45     1
   6   629   187   196     7   102   122   304     0    40     1

I'm trying to to match the lines starting with "s+number" (s0,s1,s2,...s9) and save the values between "" in a list so I can then use this list for naming the columns.

list <- c("Structure", "Coil","B-Sheet", ..., "Chain_Separato") names(data) <- list

The problem is that I can't match the single words but only the entire lines.

grep('s\\d\\s[a-z]{6}\\s\"([A-z-9]+)\"',readLines("file.xvg"),perl=T,value=T)

[1] "# s0 legend \"Structure\""       "# s1 legend \"Coil\""  
[3] "# s2 legend \"B-Sheet\""         "# s3 legend \"B-Bridge\""
[5] "# s4 legend \"Bend\""            "# s5 legend \"Turn\""
[7] "# s6 legend \"A-Helix\""         "# s9 legend \"Chain_Separator\""

I tried several regex, like '# s[0-9] [a-z]+ "([A-z-9]+)"', all working in perl but in R I'm always matching the entire line and not the word.

Isn't the () used to capture the value? What am I doing wrong?

Upvotes: 2

Views: 228

Answers (3)

Rich Scriven
Rich Scriven

Reputation: 99331

You can use a system command in fread(). For example, on a file named "file.txt" you can do

library(data.table)
fread("grep '^# s[0-9]\\+' file.txt", header = FALSE, select = 4)[[1]]
#  [1] "Structure"       "Coil"            "B-Sheet"        
#  [4] "B-Bridge"        "Bend"            "Turn"           
#  [7] "A-Helix"         "5-Helix"         "3-Helix"        
# [10] "Chain_Separator"

Note: This uses data.table dev version 1.9.5

Basically the area you're looking for in the text has four columns. ^# s[0-9]\\+ looks for lines that begin with # and then a space, then s, then any number of digits. select = 4 takes the last column, and [[1]] drops it down from a single column data table into a character vector.

Thanks to @BrodieG for help with the regex.

Upvotes: 2

akrun
akrun

Reputation: 887128

If you use linux, awk commands can be combined with read.table using pipe

read.table(pipe("awk 'BEGIN {FS=\" \"}/# s[0-9]/  { print$4 }' fra.txt"),
          stringsAsFactors=FALSE)$V1
 # [1] "Structure"       "Coil"            "B-Sheet"         "B-Bridge"  
 # [5] "Bend"            "Turn"            "A-Helix"         "5-Helix"    
 # [9] "3-Helix"         "Chain_Separator"

The above command also works with fread

fread("awk 'BEGIN {FS=\" \"}/# s[0-9]/  { print$4 }' fra.txt", 
                 header=FALSE)$V1

Upvotes: 1

Colonel Beauvel
Colonel Beauvel

Reputation: 31171

You can do this:

conn = file(fileName,open="r")
lines=readLines(conn)

lst = Filter(function(u) grepl('^# s[0-9]+', u), lines)
result = gsub('.*\"(.*)\".*','\\1',lst)

close(conn)

#> result
#[1] "Structure"       "Coil"            "B-Sheet"         "B-Bridge"        "Bend"            "Turn"            "A-Helix"         "5-Helix"        
#[9] "3-Helix"         "Chain_Separator"

Upvotes: 2

Related Questions