RNA-seq stranded coverage and junctions generated by STAR 2.6.1b

Coverages were normalized using RNA-seq log-ratio normalization factors.
Afterward, coverages were combined to the mean coverage for each experiment group.

Notes on the STAR index used for alignment:

• All samples were aligned to the same STAR index.
• The STAR index used the mouse mm39 genome and Gencode vM33 comprehensive gene annotations.
• The STAR index contains two important features:
1. A custom chromosome chrAdam19TdT was added which contains the full mutant Adam19 construct Adam19_TdTomato obtained by Sanger sequencing.
2. The wildtype Adam19 region region on chr11 not masked ("unmasked").
• Therefore two key points:
1. The STAR index should therefore allow competitive alignment of RNA-seq reads between the wildtype and mutant Adam19 exon regions.
2. This STAR index differs from the "Adam19 coverage" tracks, which had only permitted alignment of Adam19 wildtype sample RNA-seq to the wildtype Adam19 region on chr11; and only permitted alignment of mutant Adam19 sample RNA-seq to the mutant Adam19_TdTomato construct region on chrAdam19TdT.

Observations:

• All Adam19 wildtype and mutant samples show coverage on both Adam19 loci:
• the wildtype Adam19 region on chr11, and
• the region containing Adam19_TdTomato on chrAdam19TdT.
• In Adam19-wildtype samples:
• Roughly 90% of reads are assigned to chr11. This bias appears to be a choice by the STAR alignment algorithm.
• In Adam19-mutant samples:
• The majority of coverage on chrAdam19TdT is seen near the TdTomato exon which replaced wildtype exons 6 and 7 in Adam19_TdTomato.
• However, upstream from the TdTomato domain, RNA-seq reads were preferentially aligned to the wildtype Adam19 region on chr11. This bias seems consistent with the STAR alignment bias toward wildtype Adam19 region on chr11, in RNA-seq fragments which do not contain custom sequence data from the TdTomato region.
• The visual coverage represented by this STAR alignment strategy was not consistent with transcript quantitation performed by Salmon in any samples.
• Salmon uses the overall evidence to support an expectation-maximization strategy to attribute RNA-seq reads to transcript isoforms to estimate transcript abundance.
• The Salmon tool was provided all Adam19 wildtype and mutant transcripts (in addition to all Gencode vM33 comprehensive transcripts), and was freely able to attribute RNA-seq reads to any supported transcripts from that superset.
• Salmon attributed all Adam19-wildtype reads to wildtype Adam19 transcripts, and nearly all* Adam19-mutant reads to mutant Adam19 transcripts. (* >99.9%)
• The recommendation is to refer to "Adam19 coverage" as a visual representation of RNA-seq evidence, consistent with Salmon transcript isoform quantitation.