Back

Metacell projection for interpretable and quantitative use of transcriptional atlases

Ben-Kiki, O.; Bercovich, A.; Lifshitz, A.; Tanay, A.

2022-12-02 bioinformatics
10.1101/2022.12.01.518678 bioRxiv
Show abstract

We describe MCProj - an algorithm for analyzing query scRNA-seq data by projections over reference single cell atlases. We represent the reference as a manifold consisting of annotated metacell gene expression distributions. We then infer query metacells as mixtures of atlas distributions while correcting for technology-specific gene biases. This approach distinguishes and tags query cells that are consistent with existing atlas states from novel or artifactual behaviors that are not observed in the atlas. It also identifies significant expression differences observed in query states that are mapped coherently onto the atlas. We showcase MCProj functionality by analyzing blood gene expression from multiple sources and technologies, suggesting it as a method of choice for scRNA-seq analysis following extensive cell atlas projects.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

1
Genome Biology
○ 637 papers in training set
Top 0.6%
14.5%
Similar papers in this journal
2
Bioinformatics
◐ 1204 papers in training set
Top 2%
12.0%
Similar papers in this journal
3
Nature Methods
○ 385 papers in training set
Top 0.9%
10.6%
Similar papers in this journal
4
Nature Communications
○ 5641 papers in training set
Top 23%
7.0%
Similar papers in this journal
5
Nucleic Acids Research
◐ 1281 papers in training set
Top 3%
6.0%
50% of probability mass above
6
Cell Reports Methods
○ 165 papers in training set
Top 0.2%
6.0%
Similar papers in this journal
7
PLOS Computational Biology
● 1863 papers in training set
Top 8%
4.7%
8
Cell Systems
○ 201 papers in training set
Top 1%
4.2%
9
Nature Biotechnology
○ 172 papers in training set
Top 1%
3.9%
10
NAR Genomics and Bioinformatics
◐ 242 papers in training set
Top 1%
3.9%
11
Briefings in Bioinformatics
◐ 354 papers in training set
Top 3%
2.5%
12
iScience
○ 1154 papers in training set
Top 12%
2.3%
13
BMC Bioinformatics
○ 457 papers in training set
Top 3%
2.3%
14
Cell Genomics
○ 172 papers in training set
Top 2%
2.0%
15
eLife
● 5828 papers in training set
Top 49%
1.8%
16
Genome Research
● 468 papers in training set
Top 3%
1.8%
17
Proceedings of the National Academy of Sciences
● 2444 papers in training set
Top 36%
1.1%
18
Life Science Alliance
● 285 papers in training set
Top 6%
1.1%
19
Development
● 497 papers in training set
Top 4%
1.0%
20
Patterns
○ 78 papers in training set
Top 2%
1.0%
21
Bioinformatics Advances
◐ 203 papers in training set
Top 4%
1.0%
22
Science Advances
● 1243 papers in training set
Top 35%
0.6%
23
Communications Biology
○ 993 papers in training set
Top 37%
0.6%

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.