Hello,
I have hundreds of taxonomic annotation files. The 4th column of the files is the taxonomic rank. The first few rows of the files look like this:
6.46 387327 387327 U 0 unclassified
93.54 5610481 488 R 1 root
93.53 5609584 75743 R1 2 d__Bacteria
43.31 2597449 11790 R2 18 p__Actinobacteriota
23.04 1382144 149 R3 19 c__Actinobacteria
22.98 1378342 590 R4 20 o__Actinomycetales
22.55 1352503 35 R5 21 f__Bifidobacteriaceae
22.54 1352180 54264 R6 22 g__Bifidobacterium
9.43 565635 565635 R7 1797 s__Bifidobacterium adolescentis
R1=Domain, R2=Phylum, R3=Class, R4=Order, R5=Family, R6=Genus, and R7=Species.
I want to change the R[2-7] with the uppercase first letter of the respective taxonomic order. Additionally, I also want to remove the double underscores and their prefixes before the taxonomic names to make it usable for another tool. The desired output should look like this:
6.46 387327 387327 U 0 unclassified
93.54 5610481 488 R 1 root
93.53 5609584 75743 R1 2 Bacteria
43.31 2597449 11790 P 18 Actinobacteriota
23.04 1382144 149 C 19 Actinobacteria
22.98 1378342 590 O 20 Actinomycetales
22.55 1352503 35 F 21 Bifidobacteriaceae
22.54 1352180 54264 G 22 Bifidobacterium
9.43 565635 565635 S 1797 Bifidobacterium adolescentis
Please note that Bacteria (domain) would still have the R1 value. Since I have hundreds of such files, it's quite difficult to make the changes in excel or any text editor.
Could you please suggest a better option?
Many thanks for your time and help!
If your file is called input.txt, you can use:
sed '/R[2-7]/{h;s/.*\([a-z]\)__.*/\u\1/g;x;G;s/\n/ /g};s/[a-z]__//g' < input.txt | awk '{if($4 ~ /R[2-7]/) {$4=$NF;$NF=""; NF--; print} else {print $0}}' > output.txt
It's just the quick and dirty approach - the purists would probably have found a way to do it entirely in either sed or awk.
What is being done the sed part:
/R[2-7]/ matches lines that contain this pattern, so the following operations are restricted to those lines:h copies the pattern space to the hold space, while s/.*\([a-z]\)__.*/\u\1/g extracts the letter that precedes __ and converts it to uppercase. x;G;s/\n/ /g switches hold and pattern space, appends the hold space to the pattern space and replaces an introduced newline character. s/[a-z]__//g replaces the prefix of your species names. The result is a file that has an extra column at the end of the lines where we want to preserve the information from the prefix. Now awk is used, because it has a good support to restrict actions to specific columns.
if($4 ~ /R[2-7]/) is a condition that checks if the fourth column is anything between R2-R7. If no, the line is printed as it is with print $0{$4=$NF;$NF=""; NF--; print}, the fourth column is replaced with the contents of the extra column we created with sed. NF is a default variable that corresponds to the number of fields, so $NF always denotes the last field. Subsequently, that temporary extra field $NF=""; NF-- is dropped again, and the line is printed.I hope, this is what you wanted. In any case, you can customize the command to your needs of course.
PS: If you have a Mac, you need to install gsed, since Mac's default doesn't support \u for the case conversion. Alternatively, put $4=toupper($NF) in the awk part.
Matthias Zepper thank you very much for the solution and explaining it to me. I tried it on the Linux system. The command line is doing its job, but after the conversion (i.e., R# to P/C/O..etc.) 3rd row onwards all columns are becoming space separated. Is there a way to use tab as a column separator (3rd row onwards)?
Many thanks!
Yes. You can specify the output separator in awk with OFS. Add a BEGIN statement to the awk command like so:
awk 'BEGIN{OFS="\t"}; ...}' The rest of the command is unchanged.
Thanks a lot, Matthias Zepper, it solved my problem.
Log in to answer this question.