Figure 2. Flowchart of BRCA1 variant selection for exonic
SRE mapping and evaluation of SRE predictor performance.
†BRCA1 variants were from saturation genome
editing experiments (SGE) of Findlay et al. (2018) covering exons 2, 3,
5, 6 and 16-24. Cut-off scores selected for determining function class
and mRNA depletion are drawn from their Supplementary Table 1. For our
selection of negative controls, we set a conservative RNA score cut-off
≥ 0 for variants not depleted in mRNA, as some variants experimentally
confirmed to have weak to moderate splicing effects (from literature and
unpublished splicing assay data) were observed to have RNA scores
between -0.5 and -2 (data not shown).
ǂ Selection of variants outside donor and acceptor
splice site motifs and prediction of de novo splice site gain
were done through MES-based Variant Effect Predictor plugin using the
thresholds and decision flowchart described in Shamsani et al. (2018).
Figure 3. Map of putative exonic SREs (blue) in BRCA1 as
inferred from analysis of published functional data of Findlay et al.
(2018) and Human Splicing Finder SRE predictions. Findlay et al. (2018)
assessed exons 2, 3, 5, 6 and 16-24. Exons 2, 3, and 19-22 did not
harbor variants that passed the selection criteria described in Figure
2. The cDNA positions are indicated in the map, and the locations of 33
putative SRE-disrupting variants are bolded and underlined. Longer blue
regions reflect several overlapping SRE sequences that extend across the
exon. The grey area at the 5’ end of exon 16 represents a region without
functional data from Findlay et al. (2018).
Figure 4. Prioritization model to identify
SRE-disrupting variants for splicing assay. For exonic variants outside
of the donor and acceptor splice site motifs, the first step is to
filter out variants that are predicted to lead to de novo splice
site gain using the MES algorithm. Then MES is used to determine the
native splice site scores, to identify exons most likely to harbor ESEs.
Variants in exons with low to moderate donor score (< 8.5), or
with high donor MES score (≥ 8.5) and low acceptor score
(< 6.2), are then selected as eligible for two-step SRE
analysis. Positive calls in HSF are further analyzed using
ΔHZEI to minimize the number of false positives.
Supplementary Figure 1. Map of putative exonic SREs inBRCA1 exons 5, 6, 16-18, 23 and 24. The map is based on the
combined analysis of results from the functional assay of Findlay et al.
(2018), and SRE predictions from the algorithms in Human Splicing Finder
v3.1. Variants that passed the selection criteria (see main text and
Figure 2) are shown. Putative SRE locations are highlighted in blue.
cDNA positions in green font indicate locations of negative control
variants outside of putative SREs with false positive HSF SRE
predictions. cDNA positions in red font indicate locations of negative
control variants within putative SREs but with false positive HSF SRE
predictions.
Supplementary Figure 2. Snapshot of Human Splicing Finder
results webpage for MLH1 LRG_216t1:c.1559-1732A>T.The Information Theory model of Caminsky et al. (2016) predicted that
LRG_216t1:c.1559-1732A>T creates a new acceptor and
activates a pseudoexon due to the presence of a downstream pre-existing
cryptic donor, but our HSF analysis revealed a s trong putative
exonic splicing silencer (PESS) cluster within 30 nucleotides upstream
of the donor site of the pseudoexon (encircled in red), which
potentially inactivates this cryptic donor.