In my point of view since the bigger disparity is only at the begining and is corrected within 4 bases it souldnt be much problem, i dont want to filter much and lose tha characteristicas of the sequence.
I Dont know if this is really considered a major error, can someone help me?
Hello @Pedro_Fernandes_de_S
From a quick guess, your reads appear to have 12-13 bases of sequencing artifact at the start positions. To be clear – the adaptor is not a part of the molecules you are sequencing. It represents left over data from the chemistry used to capture the target molecules.
This is all expected! Since leaving those adaptor bases untrimmed can impact results, removing them is a good idea. To read more about this, please see → Hands-on: NGS data logistics / NGS data logistics / Introduction to Galaxy Analyses
Another resource summary is here. → Quality Control Start Here! multQC issue and guidance? - #2 by jennaj
Then, most other protocols at the GTN training site include QA steps at the start. Find your read type and analysis domain to review these protocol-specific steps. → https://training.galaxyproject.org/
We hope this helps! 
Thanks a lot!