PAGA - partition-based graph abstraction

Mapping out the coarse-grained connectivity structures of complex manifolds (Genome Biology, 2019).

PAGA is available within Scanpy through: tl.paga | pl.paga | pl.paga_path | pl.paga_compare.

Below you find links to all central example notebooks, which also allow reproducing all main figures of the paper. If you start working with PAGA, go through blood/paul15.

notebook	system	details	reference	figure
blood/simulated	hematopoiesis	simulated	Krumsiek et al., Plos One (2011)	2a
blood/paul15	murine hematopoiesis	2,730 cells, MARS-seq	Paul et al., Cell (2015)	2b
blood/nestorowa16	murine hematopoiesis	1,654 cells, Smart-seq2	Nestorowa et al., Blood (2016)	2c
blood/dahlin18	murine hematopoiesis	44,802 cells, 10x Genomics	Dahlin et al., Blood (2018)	2d
planaria	planaria	21,612 cells	Plass et al., Science (2018)	3
zebrafish	zebrafish embryo	53,181 cells	Wagner et al., Science (2018)	4
1M_neurons	neurons	1.3 million cells, 10x Genomics	10x Genomics (2017)	S12
deep_learning	cycling Jurkat cells	30,000 single-cell images	Eulenberg et al., Nat. Commun. (2017)	S14

All supplemental figures of the paper can be reproduced based on the following table.

notebook	description	figure
connectivity_measure	connectivity measure	S1, S2, S3
robustness	robustness and multi-resolution capacity	S4, S5
comparisons/simulated_data	comparisons for simulated data	S6, S7
comparisons/paul15_monocle2	comparison Monocle 2 for Paul et al. (2015)	S8
comparisons/nestorowa16_monocle2	comparison Monocle 2 for Nestorowa et al. (2016)	S9
embedding_quality	quantifying embedding quality	S10
simulation	simulating hematopoiesis	S11
1M_neurons	neurons, 1.3 million cells, 10x Genomics, 10x Genomics (2017)	S12
blood/paul15	annotation of louvain clusters using PAGA	S13
deep_learning	cycling Jurkat cells, 30,000 single-cell images, Eulenberg et al., Nat. Commun. (2017)	S14