Splitting concatenated PE fastq to two files for respect reads

[JayM]

I have a fastq file that is read1 and read2 split, concatenated and shuffled to form 1 file. However, I need it as read1 and read2 files separate for bwa alignment, does anyone know how to do this?
I don't have the GERALD file, this processed fastq file came from a core facility so I'm stuck at this point. Can anyone help?

[swbarnes2]

You might be able to grep it out. Something like:

grep -A 3 pattern_that_is_only_in_read_1_sample_name combined_file.fq > read1.fq

[drio]

Assuming single.

$ cat ./single.fq | ruby -ne 'BEGIN{@i=0} ; @i+=1; puts $_  if @i.to_s =~ /[1234]/; @i = 0 if @i == 8' > one.fq && cat single.fq | ruby -ne 'BEGIN{@i=0} ; @i+=1; puts $_  if @i.to_s =~ /[5678]/; @i = 0 if @i == 8' > two.fq

Also, you may want to check the fastx package. It should include that feature.

[JayM]

Assuming single.

$ cat ./single.fq | ruby -ne 'BEGIN{@i=0} ; @i+=1; puts $_  if @i.to_s =~ /[1234]/; @i = 0 if @i == 8' > one.fq && cat single.fq | ruby -ne 'BEGIN{@i=0} ; @i+=1; puts $_  if @i.to_s =~ /[5678]/; @i = 0 if @i == 8' > two.fq

Also, you may want to check the fastx package. It should include that feature.

I take it 'assume single' here refers to assume single [input] file with read1 and read2.

[JayM]

You might be able to grep it out. Something like:

grep -A 3 pattern_that_is_only_in_read_1_sample_name combined_file.fq > read1.fq

But how do you grep for read1 and not read2 from a paired end fastq given that essentially the whole name is identical except one character at the end and there are millions of such scenarios in the file...?
I'm just thinking about which pattern that could be.

[JayM]

Assuming single.

$ cat ./single.fq | ruby -ne 'BEGIN{@i=0} ; @i+=1; puts $_  if @i.to_s =~ /[1234]/; @i = 0 if @i == 8' > one.fq && cat single.fq | ruby -ne 'BEGIN{@i=0} ; @i+=1; puts $_  if @i.to_s =~ /[5678]/; @i = 0 if @i == 8' > two.fq

Also, you may want to check the fastx package. It should include that feature.

Wow! Thanks, it worked and an arbitrary inspection of the respective reads seems to confirm a perfect split into read1 and read2.