How to get rid of adapter contamination in the middle of the read

Question

[mrs.hope]

Hello!
I have performed my first run on NextSeq 500 machine. I have ChIP-seq samples, pair-end reads, 40 bp long each. I am getting very weird FastQC results for my fastq files. For the first read I consistently have k-mers enrichment in the middle of the read (see pic attached), where k-mers correspond to Illumina Universal Adapter, and the second read look OK, sometimes adapters are a bit undertrimmed in the end of the read, but it is very rare. I wonder if it looks like adapter dimer contamination or not (my Bioanalyzer traces were perfect!) considering that the second read is OK. And what will be best to use to trim these adapter sequences? I tried cutadapt but it doesn't seem to improve the k-mers enrichment at all although it identifies about 1.7% of my reads as reads with adapters.

Many many thanks in advance!

[GouthamAtla]

Where is the pic ?

[h.mon]

Were your libraries prepared following standard Illumina protocols (that is, adapters are expected to be found 3')? Did you try mapping the second read from the corresponding read 1 with adapters in the middle? Did you try the -b flag for cutadapt? Also, see here and here for other comments.