A good debate over “cutadapt” vs. “fastx-toolkit”
It turns out cutadapt has a good user community. More details can be found here.
Adam’s commands:
cutadapt -m 1 -e 0 -a TGGAATTCTCGGGTGCCAAGGAACTCCAGTCACGCCAATATCTCGTATGCCGTCTTCTGCTTG -O 5 /ddn/gs1/project/nextgen/post/paules/NG-104-Paules-Cui/ES565_121204_MARMOSET_64NDWAAXX.2.3191261.single.sanger.fastq > ES565_trimmed_0mismatch.fastq & cutadapt -m 1 -e 0.056 -a TGGAATTCTCGGGTGCCAAGGAACTCCAGTCACGCCAATATCTCGTATGCCGTCTTCTGCTTG -O 5 /ddn/gs1/project/nextgen/post/paules/NG-104-Paules-Cui/ES565_121204_MARMOSET_64NDWAAXX.2.3191261.single.sanger.fastq > ES565_trimmed_1mismatch.fastq & cutadapt -m 1 -e 0.11 -a TGGAATTCTCGGGTGCCAAGGAACTCCAGTCACGCCAATATCTCGTATGCCGTCTTCTGCTTG -O 5 /ddn/gs1/project/nextgen/post/paules/NG-104-Paules-Cui/ES565_121204_MARMOSET_64NDWAAXX.2.3191261.single.sanger.fastq > ES565_trimmed_2mismatch.fastq & cutadapt -m 1 -e 0.167 -a TGGAATTCTCGGGTGCCAAGGAACTCCAGTCACGCCAATATCTCGTATGCCGTCTTCTGCTTG -O 5 /ddn/gs1/project/nextgen/post/paules/NG-104-Paules-Cui/ES565_121204_MARMOSET_64NDWAAXX.2.3191261.single.sanger.fastq > ES565_trimmed_3mismatch.fastq &
So, based on cutadapt documentation:
-m : minimum length after trimming the adapter, default is 0 -e : number of errors divided by the length of matched base. So, if 31 bases matched with five mismatches in a 36 bp long read, it is 5/31 ~= 0.16
A few years later, cutadapt has much comprehensive user guide