Digital Normalization with khmer

Question

[Carlos Caicedo]

I am trying to run the script "normalize-by-median.py" of the khmer software but always I got this error message: "ERROR: Unpaired reads when require_paired is set!"

My question is: Do I need to merge the two files with the forward and reverse reads in one file? or the software can use each file individually and internally merge them? If the latter is the response, how I can introduce both files to the software? (In the instructions on the web I read only the way for one file)

I ran some examples but with my reads it does not work

Thank you very much for your help

Carlos

[Brian Bushnell]

I just went to the khmer documentation and found it to be 96 pages, and extremely confusing. I also remember it being rather difficult last time I used khmer. So - I can't answer your question, but maybe someone else can.

However, I can suggest BBNorm from the BBMap package, which is very easy to use:

bbnorm.sh in1=read1.fq in2=read2.fq out1=normalized1.fq out2=normalized2.fq target=100

That will normalize two files to produce 100x coverage. In my last test, it was much faster than khmer and yielded more contiguous assemblies, but that was over two years ago and I have not used khmer since, so your mileage may very.

[Carlos Caicedo]

I have already used BBMap. It was very easy to run. It helped me to obtain more assembled transcripts and the average transcripts length was bigger. Thank you very much.