Help - trimming 5'-ends

[desmodus1984]

Hi,
I am interested in mapping RNA-seq data to a genome assembly for gene annotation and I would like your hints and suggestions and comments whether I should do it or not.
I downloaded the data and I used the following parameters for fastp

fastp --thread 64
-i ${base}_1.fastq -I ${base}_2.fastq
-o ${base}_1_trim.fastq -O ${base}_2_trim.fastq
--unpaired1 ${base}_1_unp.fastq --unpaired2 ${base}_2_unp.fastq -l 20 -x
--json ${base}.fastp.json --html ${base}.fastp.htm

I recovered pairs from at least 90% of the data, but I am unsure about the initial weird base content.
The base qualities look good - in my opinion:

I thought the software might somehow trim the data in such a way that the initial base content might look better- like using a window approach or perhaps that is the signal of an adapter and that initial part might look better.

Any suggestion if I should I add any other parameter or if I shouldn't bother about that initial part.

Thanks;