I have a Fastqc and multiqc report of bunch of bulk RNA-seq (PE) data, and each of 80 Fastqc report of both raw and trimmed (Trimmomatic) shows- overrepresented sequence of (2%-6%) with possible source as TrueSeq adapters in 6 of 160 reads (2 are both pairs, 2 are either forward only). each of the read in raw and trimmed both. Duplication sequence ~45-50% is each of the read in raw and trimmed both. clipping did not work. %GC content mean value 51% has a sharper peak than theoretical distribution in all 160 raw and 154 trimmed. (FASTQC put them 154 in Warning and 6 in failed category for gc content) Adapter content is there in raw, but trimming removed it.

Fastqc/multiqc report shows that trimming of reads caused some changes in GC contents that is evident in double spikes in 6/160 trimmed read. I am sharing the 1 out of those 6 trimmed (and corresponding raw) reads that is showing failed at gc content, Adapter content in raw

gc content in raw

gc content in trimmed

Overrepresented reads are 6.56% TruSeq Adapter, Index 6 (97% over 37bp) for this trimmed and raw both.

However, Quality score is very good lies between 30-40 median range for all 160 raw and trimmed both.

How to interpret the spikes of GC content? I have explored several fastqc interpretation by its author and related fraternity discussions but could not conclude to go ahead for mapping onwards. Should i ignore the spikes and overrepresentations since sequence quality is good? or else. I appreciate for your time and suggestions.