Upgrade IsoQuant

[lianov]

1. Upgrade IsoQuant with latest improvements to parallelization.

2. If IsoQuant's parallelization is sufficient to handle large datasets, deprecate scnanoseq's internal parallelization in this step of the workflow.

[atrull314]

Noting here that we see a decrease in the amount of features that are present in the matrices (16735 in the new version of isoquant vs 60000 in the older version). We currently think that Isoquant is prefiltering features with 0 counts, we spot checked a couple features across the files and in matrices in the current version of scnanoseq and that appears to be the case. This is also supported by the median features per cell remaning unchanged despite the lower version of features

[lianov]

Adding confirmation of our thoughts below. Indeed IsoQuant now pre-filters the data. Below I ran some brief checks on the new data (from our dev) vs the old data with filtering (min.cells = 1) and the "old" data (from scnanoseq 1.2.2) without any filtering. The old-filtered data is very close to what IsoQuant is currently reporting, so no concerns there. This was tested with the test_full profile

gene-level

Checks across raw isoquant outputs (gene-level)

New isoquant outputs

seurat_obj_new
An object of class Seurat
18653 features across 812 samples within 1 assay
Active assay: RNA (18653 features, 0 variable features)
2 layers present: counts, data

Old isoquant with min.cells = 1

seurat_obj_old_filter
An object of class Seurat
18652 features across 812 samples within 1 assay
Active assay: RNA (18652 features, 0 variable features)
2 layers present: counts, data

Old isoquant with NO filtering

seurat_obj_old_NO_filter
An object of class Seurat
63186 features across 812 samples within 1 assay
Active assay: RNA (63186 features, 0 variable features)
2 layers present: counts, data

transcript-level

Checks across raw isoquant outputs (transcript-level)

New isoquant outputs

seurat_obj_new
An object of class Seurat
35344 features across 812 samples within 1 assay
Active assay: RNA (35344 features, 0 variable features)
2 layers present: counts, data

Old isoquant with min.cells = 1

seurat_obj_old_filter
An object of class Seurat
35367 features across 812 samples within 1 assay
Active assay: RNA (35367 features, 0 variable features)
2 layers present: counts, data

Old isoquant with NO filtering

seurat_obj_old_NO_filter
An object of class Seurat
252929 features across 812 samples within 1 assay
Active assay: RNA (252929 features, 0 variable features)
2 layers present: counts, data