fastq-mcf paired end adapter trimming

[gevielr]

Hi!

I'm working on a transcriptome de novo assembly, and I'm having some difficulties removing adapter contaminants from my 100 bp PE reads. According to FASTQC, I have no more than ~1% of my reads with overrepresented adapter sequences. For ex:

90743 reads 0.4% of reads TruSeq Adapter, Index 6 (100% over 50bp)
AGATCGGAAGAGCACACGTCTGAACTCCAGTCACGCCAATATCTCGTATG
42478 reads 0.10% of reads TruSeq Adapter, Index 6 (100% over 50bp)
GATCGGAAGAGCACACGTCTGAACTCCAGTCACGCCAATATCTCGTATGC

I've been tried to remove these sequences using fastq-mcf, since this seems to work well for PE reads.

However, I keep getting way more reads removed then what FastQC is telling me is present. I've been playing around with parameters, but without much improvement. I'm realizing now that the program trims partial adapter sequences from the ends, possibly even if there are just a few base pairs that match the adapter sequence? Is this generally what adapter trimming does?What if I'm only interested in trimming out the sequences overrepresented as described in FastQC, (full 65 bp-50 bp of adapter contaminants)?

There are so many parameters for this program, and I'm not sure how to set them to remove only what I need... and now I'm not sure what exactly I'm "supposed" to be removing (full or partial adapter sequence matches...)




Cheers

[mmmm]

are the below adapters sequence (OK) for trimming of illumina 250PE DNA reads
>NexteraUniversalAdapter
CTGTCTCTTATACACATCT
>TruSeq_Read1
AGATCGGAAGAGCACACGTCTGAACTCCAGTCA
>TruSeq_Read2
AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGT
>Nextera_mate_pair_Read1
CTGTCTCTTATACACATCT
>Nextera_mate_pair_Read2
AGATGTGTATAAGAGACAG
>PolyA
AAAAAAAAAAAAAAAAAAAAAAAAAAA
>sv1
AATGATACGGCGACCACCGAGATCTACACGCCTCCCTCGCGCCATCAG
>sv2
CAAGCAGAAGACGGCATACGAGAT
>sv3_barcode
CGGTCTGCCTTGCCAGCCCGCTCAG

[mastal]

@mmmm
It depends whether the library prep was done using a TruSeq kit or a Nextera kit.

[mastal]

@gevieir
The sequences that FastQC lists as over-represented are based on the first 50 bases (5' end) in the reads.

Adapter sequences are usually found towards the ends of the reads (3' end), when the insert is shorter than the read length and so you read into the adapter sequence.

So it's not surprising that you would get many more reads trimmed.

[mmmm]

library prep was done using nexteraxt so

is there a harm to include (truseq) adapters??

also, after using fastq-mcf (to trim adapters and remove bases of low qulaity <20)- then check on FastQC, still can see bases at 3' end of lower qulaity (~10-20 Q score)??