Coud you please give me an idea of the proportion of reads that typically mapped uniquely or not (but not discarded) after mapping with STAR ?