Good morning, I have a file looking like this:
file.txt
G05829 H05037 A A*02:01:01 A*11:01:01
G05829 H05037 DRA DRA*01:01:01 DRA*01:02:02
G05829 H05037 DPB1 DPB1*04:01:01 DPB1*04:02:01
G05829 H05037 DRB3 DRB3*01:01:02 DRB3*01:01:02
G05829 H05037 B B*08:01 B*44:02
G05829 H05037 DRB1 DRB1*03:01:01 DRB1*04:01:01
G15526 H12517 B B*07:02 B*35:01
G15526 H12517 DRB5 DRB5*01:01:01 DRB5*01:01:01
G15526 H12517 DRA DRA*01:02:03 DRA*01:02:03
I need to have columns 4 and 5 in the format
A*01:01 A*01:01
DRA*01:01 DRA*01:01
(...)
So, the first letters that identify the locus, star, 2 digits, column and 2 two digits.
My problem is that not eery columns have the same length. Some will have more detailed and will have 2 or 3 colons (e.g. DPB1*01:02:02 or DQB1*49:34:01:03)while other will have only one colon (the intended output, e.g. DPA*01:01).
I have tried some different approaches but I am only able to crop from the end ( what does no work because they have different lengths), crop from the beginning (Alsop dos not work because the first identifier can be 1 letter or 3 letters and number (e.g. identifier can be 'A' or 'DPB1'). I was trying with sed, by I end up replacing all the colons. My attempts:
sed 's/\:[0-9][0-9]//g' file.txt
This crops all the colon + digits WRONG
sed 's/\:[0-9][0-9]\:[0-9][0-9]\t/\t/g' file.txt
This crops only the second column and does not account for differences on length in each column.
I need something that will:
recognises identifier (A,B,C,DPA1,DQB1), the star (*), the numbers after the start (01,02,13 (..)), first colon (:) and following digits before the next column (01,02,03 ...)
so, the desired output is something like this:
niceoutput.txt
G05829 H05037 A A*02:01 A*11:01
G05829 H05037 DRA DRA*01:01 DRA*01:02
G05829 H05037 DPB1 DPB1*04:01 DPB1*04:02
G05829 H05037 DRB3 DRB3*01:01 DRB3*01:01
G05829 H05037 DRB1 DRB1*03:01 DRB1*04:01
G05829 H05037 B B*08:01 B*44:02
G15526 H12517 B B*07:02 B*35:01
G15526 H12517 DRB5 DRB5*01:01 DRB5*01:01
G15526 H12517 DRA DRA*01:02 DRA*01:02
thank you!
This sed will give You Your desired output:
sed 's/\([A-Z]\{1,\}[0-9]*\*[0-9][0-9]:[0-9][0-9]\):[0-9][0-9]/\1/g'
Test:
$ sed 's/\([A-Z]\{1,\}[0-9]*\*[0-9][0-9]:[0-9][0-9]\):[0-9][0-9]/\1/g' file.txt > niceoutput.txt
$ cat niceoutput.txt
G05829 H05037 A A*02:01 A*11:01
G05829 H05037 DRA DRA*01:01 DRA*01:02
G05829 H05037 DPB1 DPB1*04:01 DPB1*04:02
G05829 H05037 DRB3 DRB3*01:01 DRB3*01:01
G05829 H05037 B B*08:01 B*44:02
G05829 H05037 DRB1 DRB1*03:01 DRB1*04:01
G15526 H12517 B B*07:02 B*35:01
G15526 H12517 DRB5 DRB5*01:01 DRB5*01:01
G15526 H12517 DRA DRA*01:02 DRA*01:02
However in your question You are mentioning that part :[0-9] can be n times, but You do not have that test case in Your example, if that's true You will need to change sed into this:
sed 's/\([A-Z]\{1,\}[0-9]*\*[0-9][0-9]:[0-9][0-9]\)\(:[0-9][0-9]\)*/\1/g'
Test2:
$ cat jose_testcase2.txt
DPB1*01:02:02 or DQB1*49:34:01:03
DXX*05:05
$ sed 's/\([A-Z]\{1,\}[0-9]*\*[0-9][0-9]:[0-9][0-9]\)\(:[0-9][0-9]\)*/\1/g' jose_testcase2.txt
DPB1*01:02 or DQB1*49:34
DXX*05:05