Replace odd lines in one text file with the corresponding odd lines in another text file using sed/awk/etc

Question

I would like to replace one line in a text file with the corresponding line in another text file.

Contents of File1:

>line1  
GGG  
>line2  
AAA  
>line3  
TTT  
>line4  
CCC  
>line5  
CAT 

Contents of File2:

>name.A
CTGG  
>name.B
GACC  
>name.C
CTGG  
>name.D
GGAA  
>name.E
GCTA

There are the same number of lines in both files. The final file should be as follows:

>name.A  
GGG   
>name.B  
AAA  
>name.C  
TTT  
>name.D  
CCC  
>name.E  
CAT 

I've searched for a solution but can't seem to find one. I did try the following:

awk 'FNR==NR {data[FNR]=$0; next} (FNR%2==1){$0=data[FNR]}1' < File1 < File2

but nothing happens.

Any help would be much appreciated!

Answer 1

$ awk 'NR==FNR{a[NR]=$0; next} {print (FNR%2 ? $0 : a[FNR])}' file1 file2
>name.A
GGG
>name.B
AAA
>name.C
TTT
>name.D
CCC
>name.E
CAT

As well as being more concise (and working :-) ), the above is more efficient than the approach you had because:

1. Its not assigning to $0 and so isn't forcing awk to resplit the record into fields
2. It's not testing a second condition (1) for every line before deciding to print or not.

Also - don't use input redirection to open files for awk as it doesn't work for multiple files (as you found out) and robs you of the ability to check the FILENAME.

Answer 2

You could do your pbm the following way by first invoking the paste utility to arrange the lines in a TAB separated manner and then call upon the sed editor to perform surgery to obtain the desired data output:

$ paste File1 File2  | sed -ne '
    s/>.*\t//p
    s/\t.*//p
 '

Output

>name.A
GGG  
>name.B
AAA  
>name.C
TTT  
>name.D
CCC  
>name.E
CAT 

Note: Posix sed does not recognize \t escape sequence to stand in for the tab char. I just used it to highlight an otw invisible char. So you place a literal tab in place of the \t.

With Perl utility, it is a one-liner:

$ perl -pe '$_=($_,$,.<STDIN>)[$.%2]' File1 < File2

Answer 3

Your awk command is basically correct - it will work if you remove the extraneous redirections and swap the file order:

$ awk 'FNR==NR {data[FNR]=$0; next} (FNR%2==1){$0=data[FNR]}1' File2 File1
>name.A
GGG  
>name.B
AAA  
>name.C
TTT  
>name.D
CCC  
>name.E
CAT 

Alternatively, with GNU sed and a shell that supports process substitution:

$ sed -e '2~2{R /dev/stdin' -e 'd}' File2 < <(sed '1~2d' File1)
>name.A
GGG  
>name.B
AAA  
>name.C
TTT  
>name.D
CCC  
>name.E
CAT 

Explanation:

• sed '1~2d' File1 delete odd lines from File1

• < <(...) feed the results via stdin

• sed -e '2~2{R /dev/stdin' -e 'd}' File2 read one line at a time from stdin and queue it to be inserted after the next even line of File2; then delete the even line

Answer 4

Try it like this, it's complicated but works. Maybe there is an easier solution.

• First separate odds and even lines:

  awk '{print>"file1-"NR%2}' file1 awk '{print>"file2-"NR%2}' file2

This will create file1-0, file1-1, file2-0, file2-1 In my case and from your example, file1-0 had data you needed from file1 and file2-1 had data you needed from file2, check output of those files bellow

[mtodoric@bulletproof test]$ cat file1-0
GCCAAACAGCTAGCTTGA
GCCAAACAGCTAGCTTGATTAATAA
GCCAAACAGCTAGCTTGATTAATAATATAA
GCCAACCAGCTAGCTTAA
GCTAACCAGCTAGCTTAA
GCTGACTAGCTAGCCAACATATTT
GCCCAACAGCTAGCCCACTCA
GCCAACCAGCTAGCTTAAT
GCCAACCAGCTAGCTTAATT

[mtodoric@bulletproof test]$ cat file2-1
Siphateles.boraxobius_complete-mitochondrion_harney-county-or.fasta
Siphateles.boraxobius_complete-mitochondrion_harney-county-or.fasta
Siphateles.alvordensis_complete-mitochondrion_harney-county-or.fasta
Rhinichthys.evermanni_complete-mitochondrion_douglas-county-or.fasta
Oregonichthys.kalawatseti_complete-mitochondrion_linn-county-or.fasta
Oncorhynchus.mykiss.irideus_complete-mitochondrion_douglas-county-or.fasta
Cottus.perplexus_complete-mitochondrion_douglas-county-or.fasta
Oncorhynchus.mykiss.irideus_complete-mitochondrion_douglas-county-or.fasta
Cottus.perplexus_complete-mitochondrion_douglas-county-or.fasta

• Now do a "oneliner" that will combine the data.

  RB=1; while read line; do echo $line >> final.txt; sed -n ${RB}p file1-0 >> final.txt; RB=$((${RB}+1)); done < file2-1

This will basically read from file2-1 (since that file had the first line from your example output), echo each line into another file (final.txt), but also, thanks to variable RB will get that line number from another file and print stdout to final.txt, and will increase the value of RB so it can increment the line on next loop.

Replace names of file2-1 and file1-0 in "one liner" to correspond your data.

Hope this helps it.

Answer 5

Tried with sed command and it worked fine

for ((i=2;i<=10;i++));do j=`sed -n ''$i'p' f1`; k=`echo $j|sed -r "s/\s+//g"`;sed -i ""$i"s/.*/"$k"/g" f2;i=$(($i+1)); done

output

>name.A
GGG
>name.B
AAA
>name.C
TTT
>name.D
CCC
>name.E
CAT