Fig 1.
Schema of the proposed PoweREST method.
When a preliminary cohort of ST data is available, PoweREST performs the power calculation based on bootstrap and P-splines fitting. When preliminary data are not available, an R Shiny app with power estimation results based on datasets from two cancer studies can be used. Created in BioRender. Shui, L. (2025) https://BioRender.com/injyq0j.
Fig 2.
Power estimation based on the IPMN dataset.
(A and B) Histologically annotated epilesional, juxtalesional, and perilesional spots are shown in an (A) LG sample and (B) HR sample. (C) The fitted power surfaces for sample sizes of 6, 8, and 10 per group within perilesional (‘Peri’) areas. The power was fitted under the constraints so that it monotonically increases with the percentage of spots where the gene is detected and the logFC in gene expression. (D) The relationships between the power and logFC when the percentage of spots detecting the gene equals 0.1, 0.2, and 0.3. Although we did not force the monotonic relationship between sample size and power, their relationship was still monotonic in the fitted results.
Fig 3.
Validation of the other two areas from the IPMN dataset.
The fitted power surfaces for slice replicate numbers of 6, 8, and 10 per group within (A) juxtalesional (‘Juxta’) areas and (B) epilesional (‘Epi’) areas.
Fig 4.
Power estimation based on the CRC dataset.
(A and B) Histologically annotated carcinoma and carcinoma border areas in (A) an MSS sample and (B) an MSI-H sample. (C) The fitted power surfaces using P-splines for DE analysis within the carcinoma border with the number of slice replicates per group being 2, 4, and 6. (D) The relationship between the power and logFC in gene expression with slice replicates from 1 to 10 for the percentage of spots with the detected gene being 0.05, 0.10, and 0.15.
Fig 5.
Local power estimation based on the CRC dataset by XGBoost.
The power is fitted by XGBoost under constraints so that it monotonically increases with the percentage of expressed spots, the logFC and the number of slice replicates. The figures are based on power values where the logFC is between 0.1 and 1 and percentage of expressed spots is between 0.05 and 0.15. (A) The fitted power surfaces for DE analysis within the carcinoma border with the number of slice replicates per group being 2, 4, and 6. (B) The relationship between the power and logFC with slice replicates ranging from 1 to 10 for the percentage of spots with the detected gene being 0.05, 0.10, and 0.15.