The gene coverage chart created by Picard CollectRnaSeqMetrics seems to be revered - the TES and the TSS are swapped.
Typically poly-A selected libraries exhibit 3' (TES) bias, but the generated chart shows that coverage is skewed toward the "0" (which I assume indicates the TSS). See attached output.chart.pdf.
For comparison, I'm attaching a similar plot generated by ngs_plot, which exhibits 3' bias as expected. See attached si_ctrl_1.genebody.avgprof.pdf.
Am I missing something?
I am running picard 2.2.1 with the command line:
java -jar $picard CollectRnaSeqMetrics \
I=$bam \
O=output.RNA_Metrics \
CHART_OUTPUT=output.chart.pdf \
REF_FLAT=$ref_flat \
STRAND=NONE \
RIBOSOMAL_INTERVALS=$rrna