Back

Deconfounded Dimension Reduction via Partial Embeddings

Chen, A. A.; Clark, K.; Dewey, B.; DuVal, A.; Pellegrini, N.; Nair, G.; Jalkh, Y.; Khalil, S.; Zurawski, J.; Calabresi, P.; Reich, D.; Bakshi, R.; Shou, H.; Shinohara, R. T.

2023-01-11 neuroscience
10.1101/2023.01.10.523448 bioRxiv
Show abstract

Dimension reduction tools preserving similarity and graph structure such as t-SNE and UMAP can capture complex biological patterns in high-dimensional data. However, these tools typically are not designed to separate effects of interest from unwanted effects due to confounders. We introduce the partial embedding (PARE) framework, which enables removal of confounders from any distance-based dimension reduction method. We then develop partial t-SNE and partial UMAP and apply these methods to genomic and neuroimaging data. Our results show that the PARE framework can remove batch effects in single-cell sequencing data as well as separate clinical and technical variability in neuroimaging measures. We demonstrate that the PARE framework extends dimension reduction methods to highlight biological patterns of interest while effectively removing confounding effects.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

1
NeuroImage
○ 903 papers in training set
Top 1%
11.6%
Similar papers in this journal
2
Imaging Neuroscience
● 282 papers in training set
Top 0.3%
11.6%
Similar papers in this journal
3
Human Brain Mapping
○ 329 papers in training set
Top 1.0%
6.5%
Similar papers in this journal
4
Biostatistics
◐ 24 papers in training set
Top 0.1%
6.1%
Similar papers in this journal
5
Patterns
○ 78 papers in training set
Top 0.2%
5.3%
6
eLife
● 5828 papers in training set
Top 24%
5.3%
7
Nature Communications
○ 5641 papers in training set
Top 28%
5.3%
Similar papers in this journal
50% of probability mass above
8
PLOS Computational Biology
● 1863 papers in training set
Top 8%
4.2%
9
Network Neuroscience
● 126 papers in training set
Top 0.5%
3.9%
10
Medical Image Analysis
○ 35 papers in training set
Top 0.2%
3.9%
11
Nature Methods
○ 385 papers in training set
Top 3%
3.2%
12
Communications Biology
○ 993 papers in training set
Top 6%
3.1%
13
Bioinformatics
◐ 1204 papers in training set
Top 6%
2.7%
14
eneuro
● 439 papers in training set
Top 4%
2.1%
15
Proceedings of the National Academy of Sciences
● 2444 papers in training set
Top 26%
1.9%
16
Journal of Neuroscience Methods
○ 122 papers in training set
Top 0.9%
1.9%
17
The Journal of Neuroscience
● 1025 papers in training set
Top 8%
1.4%
18
Nature Human Behaviour
○ 95 papers in training set
Top 2%
1.0%
19
Brain Structure and Function
○ 93 papers in training set
Top 1%
1.0%
20
Scientific Reports
○ 3612 papers in training set
Top 70%
1.0%
21
Neuron
○ 337 papers in training set
Top 5%
0.8%
22
Cerebral Cortex
◐ 396 papers in training set
Top 5%
0.8%
23
IEEE Transactions on Medical Imaging
● 21 papers in training set
Top 0.5%
0.8%
24
Biometrics
◐ 23 papers in training set
Top 0.3%
0.8%
25
Frontiers in Neuroscience
○ 256 papers in training set
Top 6%
0.8%
26
Nature Neuroscience
○ 252 papers in training set
Top 6%
0.6%
27
Nature Computational Science
○ 55 papers in training set
Top 2%
0.6%
28
PLOS ONE
● 5266 papers in training set
Top 63%
0.6%
29
Neurobiology of Language
● 29 papers in training set
Top 0.4%
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.