CeNGEN (split by batch)

100k FACS-isolated C. elegans neurons from 17 experiments sequenced on 10x Genomics. Split into train/test by experimental batch. Dimensions: 100955 cells, 22469 genes. 169 cell types (avg. 597±800 cells per cell type).

Raw Scaled

Rank	Name	Mean score	Accuracy	Accuracy	F1 score	F1 score	Macro F1 score	Macro F1 score	Memory (GB)	Runtime (min)	CPU (%)	Year	Library
1	Logistic regression (log CP10k)	0.76	0.88	0.88	0.87	0.87	0.53	0.52	1.80	17.88	1,402.20	2013	v1.1.3
2	Multilayer perceptron (log scran)	0.74	0.87	0.86	0.87	0.86	0.49	0.49	6.80	41.85	1,580.00	1990	v1.1.3
3	Seurat reference mapping (SCTransform)	0.72	0.85	0.85	0.84	0.84	0.48	0.48	856.40	33.12	119.10	2021	v4.3.0
4	XGBoost (log scran)	0.71	0.83	0.83	0.84	0.84	0.47	0.47	6.80	53.43	1,068.40	2016	v1.6.2
5	XGBoost (log CP10k)	0.71	0.83	0.83	0.83	0.83	0.46	0.46	1.90	14.43	2,167.50	2016	v1.6.2
6	K-neighbors classifier (log CP10k)	0.69	0.82	0.82	0.84	0.83	0.42	0.42	2.60	2.30	292.30	1967	v1.1.3
7	Logistic regression (log scran)	0.68	0.77	0.77	0.80	0.80	0.48	0.47	6.80	52.22	603.40	2013	v1.1.3
8	Multilayer perceptron (log CP10k)	0.68	0.83	0.83	0.84	0.83	0.37	0.37	1.60	89.72	3,727.50	1990	v1.1.3
9	K-neighbors classifier (log scran)	0.64	0.78	0.78	0.80	0.80	0.34	0.34	6.80	39.62	106.40	1967	v1.1.3
10	scANVI (All genes)	0.48	0.65	0.64	0.62	0.61	0.20	0.20	11.60	166.77	2,504.50	2021	v0.20.0
11	scANVI (Seurat v3 2000 HVG)	0.22	0.27	0.26	0.23	0.21	0.18	0.18	20.50	177.30	1,772.90	2021	v0.20.0
12	scArches+scANVI (All genes)	0.17	0.25	0.24	0.20	0.18	0.09	0.09	15.70	118.40	2,769.00	2021	v0.20.0
13	scArches+scANVI (Seurat v3 2000 HVG)	0.04	0.08	0.07	0.04	0.02	0.04	0.04	17.90	68.90	2,431.20	2021	v0.20.0
14	Majority Vote	-0.00	0.02	0.01	0.00	-0.02	0.00	-0.00	0.99	0.80	21.20	2022	v0.7.0