Hi Maurizio,
are you saying that if you map --readFilesIn Read1.fastq Read2.fastq, you see chimeric output, but if you map with reversed mates, --readFilesIn Read2.fastq Read1.fastq, there is no chimeras detected?
This should not happen, only the strand of the alignments should revert, and there might be small changes in a very small % of mappings.
If you can send me a minimalistic example of this problem, I will check it.
Cheers
Alex