Hey @L_Zh,
So the sequencing core is performing some sort of quality filtering and trimming prior to handing you the data?
Based on that R1 sequence, it looks like the issue is that you have some reverse primers contained within the sequences. Is that correct? And due to the quality trimming these primers do not always appear at the same position.
If that's the case, you should probably use external tools to join reads and trim out primers prior to importing to QIIME2. We hope to support both read joining and cutadapt trimming soon, but these functions are not yet available in QIIME2.
Once you have joined/trimmed reads, you should be able to import these reads into QIIME2. Since you already demultiplexed these reads, you will probably want to use a fastq manifest file to import these files. Instructions in that tutorial will guide you through those steps.
I hope that helps!