Cfrtjryk

I have a csv file .Here is how it looks :

chr22,  Position ,  A     ,       B ,      C ,   D , E

22   ,  16050115 ,  0|0:404     ,  0     , 0    ,  0  ,    1|1:5

22    ,16050213    ,0|0:403  ,     0    ,  0   ,    0  ,   3|4:6

22 , 16050607  ,    1|0:340     , 1|1:3,  0   ,    0  ,    1|1:3

22   16050737 ,    0|0:402   ,   3|0:4 , 0|7:23 , 0  ,      0 

22 16050783  ,   0|0:404 ,       2|2:5 , 0|1:31 , 1|0:51,   0

0|0:404 means the count of 0|0 is 404 and so on. I would like to extract only values of 0|0,1|0,0|1 and 1:1
The output should be :

chr22, Position, A ,B, C ,D, E

22 ,16050115, 0|0:404 ,0 ,0 ,0 ,1|1:5

22 ,16050213, 0|0:403, 0 ,0, 0, 0

22 ,16050607, 1|0:340, 1|1:3, 0 ,0 ,1|1:3

22 ,16050737, 0|0:402, 0 ,0 ,0, 0

22 ,16050783, 0|0:404 ,0 ,0|1:31 ,1|0:51, 0

I tried grep -e '0|0:' -e '1|1:' -e '0|1:' -e '1|1 /path/file.csv doesn't seem to work.

Original command lines

I think the following command line will do it with a comma separated file by skipping 'offending' lines,

grep -v -e ', *[2-9]|' -e '|[2-9]:' file.csv

You can check with a small file with color,

grep -v -e ', *[2-9]|' -e '|[2-9]:' file.csv |grep --color '.|.'

and which lines are found by grep (excluded in the previous command lines by the option -v),

grep -e ', *[2-9]|' -e '|[2-9]:' file.csv |grep --color '.|.'

The following command line will do it with a tab separated file,

grep -v -e 't[2-9]|' -e '|[2-9]:' Reddy.tab

or if you want a slightly looser check, the following command line can work,

grep -v -e '[2-9]|' -e '|[2-9]:' Reddy.tab file.csv

Command line in response to a comment by the OP

I think the following command line will do what the OP wants with comma separated files as well as tab separated files,

sed -e 's/[2-9]|[0-9]:[0-9]*/NA/' -e 's/[0-9]|[2-9]:[0-9]*/NA/' file.csv

Not sure if grep is a requirement for this case, but using perl, you could do something like this to remove the unwanted items:

perl -pe 's/, ?[2-9]d*|d+:d+//g; s/, ?d+|[2-9]d*:d+//g' /path/to/file.csv


I would try

grep -e '[01]|[01]:' 

where | need to be escaped, as 0|1 means either 0 or 1.

note that

grep -e '0|0:' -e '1|1:' -e '0|1:' -e '1|1:'

match the same.

filtering on a column :

awk -F, '$3 ~/[01]|[01]:/ '

where $3 is third column.

filtering column 3 and above

awk -F, '{printf "%s,%s",$1,$2 ; 

      for(i=3;i<=NF;i++) 

        if ( $i  ~ /[01]|[01]:/) 

           printf ",%s",$i ; 

      printf "n" }' 

that can be one lined.

• note that | in this case, need to be escaped.


• 
  

  note also that first line is filtered out

  
  
  

  chr22,Position
  
  22,16050115,0|0:404,1|1:5
  
  22,16050213,0|0:403
  
  22,16050607,1|0:340,1|1:3,1|1:3
  
  22,16050737,0|0:402
  
  22,16050783,0|0:404,0|1:31,1|0:51