NovoSpaRc: flexible spatial reconstruction of single-cell gene expression with optimal transport

Moriel, Noa; Senel, Enes; Friedman, Nir; Rajewsky, Nikolaus; Karaiskos, Nikos; Nitzan, Mor

doi:10.1038/s41596-021-00573-7

Cited by 79 publications

(65 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We turn our attention to an application of map estimation using real-world data, where practitioners may not have a priori knowledge of a map even existing between the source and target measures. Such an example arises in (Demetci et al, 2021;Moriel et al, 2021;Schiebinger et al, 2019), where the task is to infer cellular evolution from population measurements. The original data is temporal, where cell measurements are taken across 18 days, and each sampled data point consists of over 1000 gene expressions.…”

Section: Application: Predicting Trajectories Of Genomesmentioning

confidence: 99%

See 1 more Smart Citation

Debiaser Beware: Pitfalls of Centering Regularized Transport Maps

Pooladian¹,

Cuturi²,

Niles-Weed³

2022

Preprint

View full text Add to dashboard Cite

Estimating optimal transport (OT) maps (a.k.a. Monge maps) between two measures P and Q is a problem fraught with computational and statistical challenges. A promising approach lies in using the dual potential functions obtained when solving an entropy-regularized OT problem between samples Pn and Qn, which can be used to recover an approximately optimal map. The negentropy penalization in that scheme introduces, however, an estimation bias that grows with the regularization strength. A well-known remedy to debias such estimates, which has gained wide popularity among practitioners of regularized OT, is to center them, by subtracting auxiliary problems involving Pn and itself, as well as Qn and itself. We do prove that, under favorable conditions on P and Q, debiasing can yield better approximations to the Monge map. However, and perhaps surprisingly, we present a few cases in which debiasing is provably detrimental in a statistical sense, notably when the regularization strength is large or the number of samples is small. These claims are validated experimentally on synthetic and real datasets, and should reopen the debate on whether debiasing is needed when using entropic optimal transport.

show abstract

Section: Application: Predicting Trajectories Of Genomesmentioning

confidence: 99%

“…. , Y n ∼ Q, estimating such maps is even more challenging, yet increasingly relevant when, for instance, trying to infer cellular evolution from population measurements (Demetci et al, 2021;Moriel et al, 2021;Schiebinger et al, 2019;Yang et al, 2020).…”

Section: Introductionmentioning

confidence: 99%

Debiaser Beware: Pitfalls of Centering Regularized Transport Maps

Pooladian¹,

Cuturi²,

Niles-Weed³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…In situ hybridizations of representative marker genes were obtained from the Berkeley Drosophila Genome Project [39][40][41]. Colors representing Leiden clusters were projected onto a virtual embryo using novoSpaRc [42,43].…”

Section: Data Processing and Analysismentioning

confidence: 99%

“…(i) Projection of nuclei onto a virtual embryo labeled by the Leiden cluster as colored in Figure 1a. Virtual in situ hybridizations and projection of clusters onto a virtual embryo were generated using novoSpaRc [42,43].…”

Section: Figure 2 Leiden Clusters Correspond To Spatial Regions Within the Embryo (A-h)mentioning

confidence: 99%

Single-nucleus RNA-sequencing in pre-cellularizationDrosophila melanogasterembryos

Albright

Stadler

2021

Preprint

View full text Add to dashboard Cite

Our current understanding of the regulation of gene expression in the early Drosophila melanogaster embryo comes from observations of a few genes at a time, as with in situ hybridizations, or observation of gene expression levels without regards to patterning, as with RNA-sequencing. Single-nucleus RNA-sequencing however, has the potential to provide new insights into the regulation of gene expression for many genes at once while simultaneously retaining information regarding the position of each nucleus prior to dissociation based on patterned gene expression. In order to establish the practicality of single-nucleus RNA sequencing in the context of a real biological question, here we look at the difference in gene expression between control and an insulator protein, dCTCF, maternal null embryos during zygotic genome activation at nuclear cycle 14. We find that early embryonic nuclei can be grouped into distinct clusters according to gene expression. From both virtual and published in situ hybridizations, we also find that these clusters correspond to spatial regions of the embryo. Lastly, we present multiple examples of differential gene expression between control and maternal CTCF null nuclei in one or more clusters, but not in bulk when grouping expression across all nuclei. These results highlight the potential for single-nucleus RNA-sequencing to reveal new insights into the regulation of gene expression in the early Drosophila melanogaster embryo.

show abstract

“…Consequently, mapping single cells onto a spatial coordinate system is equivalent to performing appropriate dimensional reduction on the entire gene expression matrix or some meaningful subset of genes (Karaiskos et al 2017, Nowotschin et al 2019). Specific approaches for inferring the physical space have varied in complexity, from applications of classical dimensionality reduction techniques such as Principal Component Analysis (Durruthy-Durruthy et al 2014, Mori et al 2019, Ren et al 2020) or latent variable models (Verma and Engelhardt 2020) to approaches using concepts from optimal transport, such as novoSpaRc (Moriel et al 2021) and SpaOTsc (Cang and Nie 2020). In all cases, a key limitation of these approaches is the underlying assumption: cells with similar gene expression profiles are proximal in space.…”

Section: Introductionmentioning

confidence: 99%

Supervised spatial inference of dissociated single-cell data with SageNet

Heidari

Lohoff

Tyser

et al. 2022

Preprint

View full text Add to dashboard Cite

Spatially-resolved transcriptomics uncovers patterns of gene expression at supercellular, cellular, or subcellular resolution, providing insights into spatially variable cellular functions, diffusible morphogens, and cell-cell interactions. However, for practical reasons, multiplexed single cell RNA-sequencing remains the most widely used technology for profiling transcriptomes of single cells, especially in the context of large-scale anatomical atlassing. Devising techniques to accurately predict the latent physical positions as well as the latent cell-cell proximities of such dissociated cells, represents an exciting and new challenge. Most of the current approaches rely on an ‘autocorrelation’ assumption, i.e., cells with similar transcriptomic profiles are located close to each other in physical space and vice versa. However, this is not always the case in native biological contexts due to complex morphological and functional patterning. To address this challenge, we developed SageNet, a graph neural network approach that spatially reconstructs dissociated single cell data using one or more spatial references. SageNet first estimates a gene-gene interaction network from a reference spatial dataset. This informs the structure of the graph on which the graph neural network is trained to predict the region of dissociated cells. Finally, SageNet produces a low-dimensional embedding of the query dataset, corresponding to the reconstructed spatial coordinates of the dissociated tissue. Furthermore, SageNet reveals spatially informative genes by extracting the most important features from the neural network model. We demonstrate the utility and robust performance of SageNet using molecule-resolved seqFISH and spot-based Spatial Transcriptomics reference datasets as well as dissociated single-cell data, across multiple biological contexts. SageNet is provided as an open-source python software package at https://github.com/MarioniLab/SageNet.

show abstract

NovoSpaRc: flexible spatial reconstruction of single-cell gene expression with optimal transport

Abstract: This protocol describes novoSpaRc, a computational pipeline for de novo reconstruction of spatial gene expression from single-cell RNA sequencing with the potential to incorporate spatial atlas data to improve the reconstruction.

Cited by 79 publications

References 50 publications

Debiaser Beware: Pitfalls of Centering Regularized Transport Maps

Debiaser Beware: Pitfalls of Centering Regularized Transport Maps

Single-nucleus RNA-sequencing in pre-cellularizationDrosophila melanogasterembryos

Supervised spatial inference of dissociated single-cell data with SageNet

Contact Info

Product

Resources

About