Changelog¶
v1.0¶
- Main components of
scallop
: sl.Scallop
andsl.Bootstrap
objects.tl:
sl.tl.getScore()
.pl:
sl.pl.plotScore()
.datasets:
sl.datasets.delete_datasets()
andsl.datasets.joost2016()
.
v1.0.1¶
Fixed pip dependencies.
v1.0.2¶
Corrected calculation of freqScore
to range between 0 and 1.
v1.0.3¶
Fixed minor issues.
v1.0.4¶
Corrected scallop clustering use_weights
parameter inferred from scanpy’s.
v1.1.0¶
Added logging support.
Implemented parallelization of
sl.tl.getScore()
.Added
louvain
to clustering algorithms.Added 10x mouse brain and heart 10k datasets.
Remapping of unknown clusters.
v1.1.1¶
Corrected error on sampling for bootstrap part of sl.tl.getScore()
.
When bootstrapping, a set of random indices are selected. If frac_cells=1
then all indices are selected. However, those indices were randomized, and
even with the same seed
value, clustering would yield different results.
Fixing the issue implies to sort the selected indices, regardless of frac_cells
selected.
v1.2.0¶
Implemented
force
option forsl.tl.getScore()
.Applied some assertions on
sl.tl.getScore()
.Added dictionary of arguments to add seed to parallel calculation of
bt_matrix
.Added
seed_in_bt
as argument forsl.tl.getScore()
.Corrected
seed_in_bt
to solveNone
problem from leiden. In some datasets, if seed is None, the random generation algorithm only considers 6-or-so different seeds.
v1.3.0¶
Ensure compatibility with latest Scanpy AnnData structure (v1.8).
Complete the documentation.
Ensure compatibility with ray latest version.
Minor fix in nan management during freqScore calculation.