Selective Inference for Hierarchical Clustering

Gao, Lucy L.; Bien, Jacob; Witten, Daniela

doi:10.1080/01621459.2022.2116331

Cited by 49 publications

(62 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, we judged that the data-driven spatial domains identified from BayesSpace were not sufficiently reliable to use for the downstream analyses, and instead proceeded with the histology-driven manual annotations for all further analyses. In addition, we note that using the manual annotations avoids potential issues due to inflated false discoveries resulting from circularity when performing differential gene expression testing between sets of cells or spots defined by unsupervised clustering, when the same genes are used for both clustering and differential testing [37]. Next, in addition to the manually annotated LC regions, we also manually annotated a set of individual spots that overlapped with NE neuron cell bodies identified within the LC regions, based on pigmentation, cell size, and morphology from the H&E histology images ( Supplementary Figure 6A ).…”

Section: Resultsmentioning

confidence: 99%

The gene expression landscape of the human locus coeruleus revealed by single-nucleus and spatially-resolved transcriptomics

Weber

Divecha

Tran

et al. 2022

Preprint

View full text Add to dashboard Cite

Norepinephrine (NE) neurons in the locus coeruleus (LC) project widely throughout the central nervous system, playing critical roles in arousal and mood, as well as various components of cognition including attention, learning, and memory. The LC-NE system is also implicated in multiple neurological and neuropsychiatric disorders. Importantly, LC-NE neurons are highly sensitive to degeneration in both Alzheimer's and Parkinson's disease. Despite the clinical importance of the brain region and the prominent role of LC-NE neurons in a variety of brain and behavioral functions, a detailed molecular characterization of the LC is lacking. Here, we used a combination of spatially-resolved transcriptomics and single-nucleus RNA-sequencing to characterize the molecular landscape of the LC region and the transcriptomic profile of LC-NE neurons in the human brain. We provide a freely accessible resource of these data in web-accessible formats.

show abstract

Section: Resultsmentioning

confidence: 99%

The gene expression landscape of the human locus coeruleus revealed by single-nucleus and spatially-resolved transcriptomics

Weber

Divecha

Tran

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Supervised feature selection, which involves filtering features that are based on association with the outcome, is another form of preprocessing that can cause leakage when performed outside cross validation. More broadly, recent work on post selection inference highlights the problem of performing statistical analyses such as differential expression after clustering 72 , even if the clusters were defined on independent datasets 73 .…”

Section: Pitfall 4: Leaky Preprocessingmentioning

confidence: 99%

Navigating the pitfalls of applying machine learning in genomics

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Whilst our approach has been developed for changepoint problems, the general idea can be applied to other scenarios such as clustering (Gao et al, 2022;Chen and Witten, 2022) or regression tress (Neufeld et al, 2022). For example, current methods for post-selection inference after clustering are based on a test statistic that compares the mean of the cluster, and fixed the projection of the data that is orthogonal to this.…”

Section: Discussionmentioning

confidence: 99%

Improving Power by Conditioning on Less in Post-selection Inference for Changepoints

Carrington¹,

Fearnhead²

2023

Preprint

View full text Add to dashboard Cite

Post-selection inference has recently been proposed as a way of quantifying uncertainty about detected changepoints. The idea is to run a changepoint detection algorithm, and then re-use the same data to perform a test for a change near each of the detected changes. By defining the p-value for the test appropriately, so that it is conditional on the information used to choose the test, this approach will produce valid p-values. We show how to improve the power of these procedures by conditioning on less information. This gives rise to an ideal selective p-value that is intractable but can be approximated by Monte Carlo. We show that for any Monte Carlo sample size, this procedure produces valid p-values, and empirically that noticeable increase in power is possible with only very modest Monte Carlo sample sizes.Our procedure is easy to implement given existing post-selection inference methods, as we just need to generate perturbations of the data set and re-apply the post-selection method to each of these. On genomic data consisting of human GC content, our procedure increases the number of significant changepoints that are detected from e.g. 17 to 27, when compared to the method of Jewell et al. (2021).

show abstract

Selective Inference for Hierarchical Clustering

Cited by 49 publications

References 30 publications

The gene expression landscape of the human locus coeruleus revealed by single-nucleus and spatially-resolved transcriptomics

The gene expression landscape of the human locus coeruleus revealed by single-nucleus and spatially-resolved transcriptomics

Navigating the pitfalls of applying machine learning in genomics

Improving Power by Conditioning on Less in Post-selection Inference for Changepoints

Contact Info

Product

Resources

About