I regularly run alignment + variant calling pipelines for Illumina DNA-seq reads. Sometimes I can learn from whoever did the upstream sequencing whether there are adapters in the reads that I should look for to trim. But most of the times the person/organization who gave me the reads had no such information at all.
My usual routine is just to run FastQC to see whether there are warning for the "Adapter Content". If there's a red flag, I'll try to run some script to identify/trim the adapters. If not, I'll just directly pass the reads to the aligners.
I'm wondering would this be considered as the best practice? What does Illumina (HiSeq/MiSeq) software usually do for QC before they output the fastq files? If it's not a routine for them to trim the adapters, how does the software decide when to trim or not (based on some parameters)? Thanks in advance for any information!
I recommend always trimming adapters unless you specifically need them for some purpose. Adapter trimming is a very fast and safe step when done correctly and reduces noise in all stages downstream. FastQC will only warn you about overrepresented sequences if the amount of adapter content passes some threshold, but there will virtually always be adapter contamination, and it's not going to help you to leave it there.