Fragment is hung over the end of the transcript

[kelly-sovacool]

Error from Bahman Afsari, using a custom hg38 reference with HPV included.

Error arises during RSEM:

Fragment lh00172:623:23HWW7LT4:7:2192:47550:1813 is hung over the end of transcript 173612! It is possible that the aligner you use gave different read lengths for a same read in SAM file.
Found unknown sequence letter � at function get_base_id!
Found unknown sequence letter � at function get_base_id!

Root cause: the star alignment rules include --alignEndsProtrude 10 ConcordantPair and --peOverlapNbasesMin 10, which messes up RSEM read quantification. These params have minimal impact on long human transcripts in hg38, but for HPV's short transcripts it's the extra 10 bp hanging over the end is more noticeable.