Missed call - not clear why

[gevro]

Hi,
We did RNA-seq on a sample on which we know there is a very clear and strong splicing defect, as a positive control for the DROP pipeline.

The exon skipping is very obvious:

However, DROP did not detect it.

Here are our config parameters:

aberrantSplicing:
    run: true
    groups:
        - GTEX100
    recount: false
    longRead: false
    keepNonStandardChrs: false
    filter: true
    minExpressionInOneSample: 20
    minDeltaPsi: 0.05
    implementation: PCA
    padjCutoff: 0.1
    maxTestedDimensionProportion: 6
    genesToTest: null
    FRASER_version: "FRASER2"
    deltaPsiCutoff : 0.1
    quantileForFiltering: 0.75

Here is the call in results_gene_all.tsv, but it was not in the final FRASER output:

seqnames	start	end	width	strand	sampleID	hgncSymbol	type	pValue	psiValue	deltaPsi	counts	totalCounts	meanCounts	meanTotalCounts	nonsplitCounts	nonsplitProportion	nonsplitProportion_99quantile	annotatedJunction	pValueGene	padjustGene
**	**	**	3650	+	Sample_RNA	**	jaccard	0.032264	0.31	-0.14	146	464	123.63	128.96	59	0.13	NA	both	0.90339	1

Can you help us understand why this did not get into the final call set and how we can fix this?

Also would appreciate to know if there is a fix, how we can do that without rerunning the whole pipeline, which takes a long time.

Thanks

[vyepez88]

Hi,
Thanks for using DROP and reporting this.
It would be interesting to see the results for all 3 of the junctions involved in the exon skipping event you are focusing on here. Can you run the following command to get the results table of all the junctions of your sample of interest?

results(fds, sampleIDs = {your sample of interest},  aggregate = FALSE,  all = TRUE)

Also, from the one junction that you provided the results, it's hard for us to understand what might be happening. Can you provide the output of the plotExpectedVsObservedPsi function which can be a starting point for understanding why FRASER reports a relatively small, non-significant delta jaccard value in this case?

[gevro]

Sorry, none of these work. I'm getting these errors:

> results(fds,sampleIDs = "Sample_RNA",  aggregate = FALSE,  all = TRUE)
Sat Sep 16 16:06:31 2023: Collecting results for: psi3
Error: BiocParallel errors
  element index: 1, 2, 3
  first error: error in evaluating the argument 'x' in selecting a method for function 'rowMeans': the supplied 'dimnames' must have one list element per dimension

> plotExpectedVsObservedPsi(fds,type="psi5")
Error in normarg_dimnames(dimnames, seed_dim) : 
  the supplied 'dimnames' must have one list element per dimension