Reference

All functions

add_simba_scores()

update an SCE with simba edge scores

disc_matrix()

discretize a dense matrix

docol()

helper function

embs_to_scores()

Produce a SummarizedExperiment instance with assay elements (i,j) the simba score relating gene i to cell j

fb15k_folder()

set up folder with fb15k demo data

flatten_json()

Helper function to recursively flatten nested lists produced by Chat-GPT 3

gini()

gini coefficients from rows of a matrix

ingest_training_stats()

get pbg_metrics data frames

make_entity_schema()

produce EntitySchema

make_rel_schema()

produce RelationSchema for torchbiggraph configuration

make_triples()

make triples from a discretized SCE

make_trvate()

convert data.frame to train/test/validate subsamples

max_scores()

obtain the simba max metric for genes

nn50

embeddings for PBMC3K based on 50 epochs of training by torch-biggraph for 4780 highly variable genes on 2700 cells.

norm_scores()

obtain the simba norm metric for genes

path_to_15k_tgz()

cache and/or retrieve path to fb15k.tgz

pca_CG()

update a SingleCellExperiment with PCA (via irlba) reductions of embeddings produced with PyTorch-BigGraph

sce_to_embeddings()

produce torchbiggraph embeddings for discretized single-cell RNA-seq measures in a SingleCellExperiment

sce_to_triples()

filter a discretized SCE using getTopHVGs from scran, the produce triples (edges) (cell - weight - gene)

setup_config_schema()

make ConfigSchema

simba_barplot_df()

set up a data.frame for barplot visualization for a gene

t3k

a SingleCellExperiment for PBMC3K that includes assay elements logcounts (produced by scran::logNormCounts) and disc (produced by BiocPBG::disc_matrix)

train_eval()

run training and (optionally) evaluation

triples_to_hdf5()

use config object

triples_to_hdf5_raw()

produce HDF5 files for edges defined in tsv files of triples (left, rel, right)

umap_CG()

update a SingleCellExperiment with UMAP reductions of embeddings produced with PyTorch-BigGraph