Datasets
Seventy-one ABCA4 NCSS, 81 ABCA4 DI and 61 MYBPC3NCSS variants with functional validation were taken from Khan et
al. (ABCA4 ) and Ito et al. (MYBPC3 ) (Figure 1a,
Supplementary Table 1) (Ito et al., 2017; Khan, Cornelis, Pozo-Valero,
et al., 2020). The selection criterion for functional validation of theABCA4 variants was a 2% difference in splice score for at least
two of the Alamut programs (SpliceSiteFinder-like, MaxEntScan, NNSPLICE,
GeneSplicer and HSF) and a relative strength of at least 75% for novel
splice sites. To assess pathogenicity of putative causative ABCA4variants, splice assays were performed using midigenes as previously
described (Fadaie et al., 2019; Khan, Cornelis, Pozo-Valero, et al.,
2020; Sangermano et al., 2018). In short, midigenes contain multiple
exons and introns to create a natural genomic context for testing the
effect of variants on splicing. A construct containing the wildtype is
then compared to a construct including the mutant variant which is
introduced by site-directed mutagenesis. After independent transfection
into HEK293T cells, ABCA4 transcripts were amplified using RT-PCR
and separated on a 2% agarose gel to determine the percentage of mutant
RNA in comparison with the control line. ABCA4 variants with more
than 20% mutant RNA were classified as splice altering (Sangermano et
al., 2018). MYBPC3 variants with a smaller MaxEntScan score than
the reference nucleotide were assessed in minigenes that contained a CMV
promoter and a 500 bp oligonucleotide with the relevant intron flanked
by exon fragments (Ito et al., 2017). Computational quantification of
qPCR transcripts with a significant difference (p < 0.001,
two-sided Fisher’s exact test) between wildtype and mutant transcript
was performed, and variants with a significant difference were
classified as splice altering. Both the ABCA4 and MYBPC3datasets were aligned to the human genome reference GRCh37/hg19
assembly.