I have strand-specific RNASeq Illumina data (TrueSeq library preparation kit). After noticing adaptor contamination via FASTQC, I realized that the adaptor* sequence is always followed by polyA tail. I wonder why.

In one of the old posts, Jeremy mentiones something as "fakePolyA": C: How to interpret the kmer enrichment plot of a FastQC output

This can be seen only in /1 reads and looks like this: GATCGGAAGAGCACACGTCTGAACTCCAGTCACCGATGTATCTCGTATGCCGTCTTCTGCTTGAAAAAAAAAAAAAAAAA[here continues variable sequence that I think comes from the real sample:)]

Interestingly, almost all sequences have this A+ tail:

grep -E "GATCGGAAGAGCACACGTCTGAACTCCAGTCACCGATGTATCTCGTATGCCGTCTTCTGCTTG[A]+" R1_001.fastq | wc -l 5108

grep -E "GATCGGAAGAGCACACGTCTGAACTCCAGTCACCGATGTATCTCGTATGCCGTCTTCTGCTTG" R1_001.fastq | wc -l 5119

Any idea will be appreciated:-)

*I found adaptor sequences at this site: http://supportres.illumina.com/documents/myillumina/6378de81-c0cc-47d0-9281-724878bb1c30/2012-09-18_illuminacustomersequenceletter.pdf

EDIT: I have cut the first four characters of the fastq file and joined it with polyA (polyA tails longer than 6 bp were trimmed):

GATCAAAAAA

It seems that scores are pretty good. It's definitely not zero.

EDIT2:

Here are the actual fastq substrings. I am sorry that I wrote before that I provided the part between adaptor and polyA (which would be CTTGAAAAAA). I actually cut first 4 characters (GATC) and joined it with polyA (this results in GATCAAAAAA). This however does not effect polyA quality scores from fastq files:)