PPubMed21 Jun 2021Cited 290×

SPARK-X: non-parametric modeling enables scalable and robust detection of spatial expression patterns for large spatial transcriptomic studies

Jiaqiang Zhu, Shiquan Sun, Xiang Zhou

Abstract

Spatial transcriptomic studies are becoming increasingly common and large, posing important statistical and computational challenges for many analytic tasks. Here, we present SPARK-X, a non-parametric method for rapid and effective detection of spatially expressed genes in large spatial transcriptomic studies. SPARK-X not only produces effective type I error control and high power but also brings orders of magnitude computational savings. We apply SPARK-X to analyze three large datasets, one of which is only analyzable by SPARK-X. In these data, SPARK-X identifies many spatially expressed genes including those that are spatially expressed within the same cell type, revealing new biological insights. Supplementary Information The online version contains supplementary material available at 10.1186/s13059-021-02404-0.

A figure from SPARK-X: non-parametric modeling enables scalable and robust detection of spatial expression patterns for large spatial transcriptomic studies — fig. from the paper

§ The Valyu brief

The paper, in five minutes

Read from the full text via Valyu Contents

SPARK-X is a non‑parametric tool that rapidly and accurately detects spatial gene expression patterns in massive spatial transcriptomics datasets.

What’s new

i.Developed SPARK‑X, a non‑parametric method for detecting spatially expressed genes
ii.Achieved orders‑of‑magnitude speed and memory improvements over existing methods
iii.Applied SPARK‑X to large Slide‑seq, Slide‑seqV2, and HDST datasets, uncovering many spatial genes
iv.Provided calibrated p‑values and high statistical power

How they did it

SPARK‑X tests independence between gene expression and spatial distance using a non‑parametric covariance statistic and combines evidence across multiple distance matrices. Significance is assessed via permutations, providing calibrated p‑values. The approach was evaluated on simulated data and three large real spatial transcriptomics datasets.

What they found

Analyzed 10,000 genes across 20,000 locations in 1 min using 0.32 GB RAM (vs 1125 min/21.2 GB for SPARK‑G and 770 min/55.3 GB for SpatialDE).
In mouse cerebellum Slide‑seq data (17,729 genes, 25,551 beads), SPARK‑X identified 2,336 spatially expressed genes at 1 % FDR, ~10× more than SPARK‑G (212).
In mouse olfactory bulb HDST data (19,913 genes, 177,455 spots), SPARK‑X completed analysis in 3 min with 0.42 GB RAM, whereas SPARK‑G would need ~114 days and 2100 GB memory.
SPARK‑X detected 125 spatially variable genes across eight morphological layers, revealing distinct gene clusters.

Caveats

✕Only a limited set of distance covariance functions were explored; other functions may improve power.
✕The Poisson version of SPARK was not applied due to excessive computational requirements.
✕Validation was performed on a few published datasets, leaving broader applicability to be tested.
✕Method relies on permutation testing, which may be computationally intensive for extremely large datasets.

§ The Valyu brief

Reading the full paper and taking notes. This takes a few seconds…

§ Ask this paper

Ask a question about this paper

Valyu reads the full text and answers from what the paper actually says.

§ Related work in other archives

SPARK-X: non-parametric modeling enables scalable and robust detection of spatial expression patterns for large spatial transcriptomic studies

The paper, in five minutes

Ask a question about this paper

Similar papers in all four archives

Similar papers in all four archives

The paper, in five minutes