Change log: Analyses

Version updates to Pluto's out-of-the-box analyses

Overview

Analyses in Pluto refer to a library of containerized bioinformatics pipelines that can be enabled for various experiment types. We are constantly working to improve the product experience, and regularly ship new updates to our standard analyses. Questions about this? Feel free to contact us.

When a new version of an analysis is available, you'll see a green banner on the Analysis tab of your existing analyses notifying you that a new version is available.

After reviewing the changes made in the latest version, click the "Run analysis" button if you want to update your analysis version.

Change log

Note: The change log below does not include proprietary or customer-specific analyses. If your organization has configured its own analyses, your Pluto customer rep will communicate with you directly via email about any changes, per the terms in your signed service agreement.

Clustering analysis

v1.2 (2026-08-28)

CPM normalization is now calculated across all features in the assay data before the selected features are subset for clustering, rather than after. Previously, when CPM normalization was enabled, each sample's library size was computed only from the subset of features being clustered, so the normalized values depended on which features were selected. Selecting a directional subset — for example, all downregulated features — could bias the distribution and make those features appear less strongly regulated than they actually are. CPM values now reflect each sample's full library size and are independent of the feature selection.

Re-run an existing clustering analysis to upgrade it to v1.2; prior results remain available as originally run.

v1.1 (2026-07-22)

Clustering can now be run on a custom gene/target list. In addition to the existing option of using significant features from a differential analysis, you can now select features from a biomarker/target list or a followed list. This removes the requirement to run a differential analysis first when you already have a curated set of targets you want to cluster on.

When using the differential-features source, fold-change filtering now supports independently configurable lower and upper thresholds along with a direction (up, down, or both), giving finer control over which features are included in the clustering. The analysis Methods text also now names the specific biomarker/target list used, so the write-up reflects exactly which targets were clustered.

Re-run an existing clustering analysis to upgrade it to v1.1; prior results remain available as originally run.

Over-representation analysis (single cell)

v1.1 (2026-07-17)

Top-gene selection now ranks by absolute fold change. For 1:1 comparisons this ensures the most strongly downregulated genes (not just those nearest zero) are included for the negative-fold-change group. Re-run an existing ORA to upgrade it to v1.1; prior results remain available as originally run.

Differential expression analysis

v1.3 (2025-07-15)

Support for filtering parameters was added to differential gene expression analysis. Features can now be filtered based on a minimum raw read count threshold, as well as the minimum percentage of samples in each group (control and experimental) that meet this threshold. These parameters are independently configurable by group, enabling refined selection of features based on biological or experimental context. Learn more about feature filtering for differential analysis.

In addition, the underlying runtime environment was upgraded. The Docker image was migrated from a Conda-based setup using R 4.1 to a new container based on R 4.5.1 (rocker/r-ver). Package management is now handled via BiocManager (v3.21), with installation of core libraries including DESeq2, edgeR, data.table, and BiocParallel. These changes reduce container complexity and improve performance in distributed compute environments. No changes were made to the statistical modeling framework or existing default parameterization of DESeq2.

v1.2 (2023-11-27)

Added the ability to optionally include covariates when running differential expression analysis with DESeq2. Learn more about using covariates.

v1.1 (2022-07-27)

Added heatmap plot option, which displays either the calculated z-score, average or mean counts per million (CPM) value of the experimental and control groups.

v1.0

Differential expression analysis was implemented using the DESeq function in the DESeq2 package (v1.24.0) with parameter pAdjustMethod = "fdr".

Differential expression (continuous) analysis

v1.1 (2025-07-15)

Infrastructure upgrades were applied to improve performance and maintainability of the Differential expression (continuous) pipeline. The analysis was migrated from a Conda-based Docker image using R 4.1 to a new container based on R 4.5.1. Package installation is now performed using BiocManager (v3.21), including the use of limma and data.table. These updates reduce image complexity, improve runtime efficiency, and provide forward compatibility with updated R packages.

No modifications were made to the statistical modeling approach or interpretation of continuous expression values.

Differential peak analysis

v1.3 (2025-07-15)

Differential peak analysis was updated to include the same new filtering capabilities introduced in Differential expression analysis v1.3. Features can be excluded based on a minimum read count and a minimum proportion of samples per group meeting that threshold. These settings are independently defined for each group, providing greater flexibility in controlling for variability in sequencing depth or biological heterogeneity. Learn more about feature filtering for differential analysis.

This release also included a full environment upgrade to align with the Differential expression pipeline (v1.3). The Docker image was rebuilt using R 4.5.1 and BiocManager (v3.21) for installation of all dependencies. The runtime is now more efficient, and the container is better optimized for cloud-based execution. No changes were made to the statistical modeling framework or existing default parameterization of DESeq2.

Gene set enrichment analysis

v1.2 (2022-12-30)

Added the ability to select from multiple different methods for ranking genes. The default (and method used for all prior runs) is "fold change." The newly added options allow filtering by p-value (non-directional), signed p-value, and fold change * p-value.

v1.1 (2022-07-18)

Added the Gene column to the enrichment plot data table. No changes were made to the calculation of the enrichment plot x and y values so the plots are unaffected besides the addition of the gene name in the hover text.

v1.0

Gene set enrichment analysis (GSEA) was implemented using the fgseaMultilevel function in the fgsea package (v1.16.0). GSEA produces data for the enrichment plot as well as a gene sets summary table.

Image

v1.0

Image upload was implemented. No transformations to the actual image are run, the user can upload an image, write in the methods, and tag one or more targets that are present in the image.

Principal components analysis

v1.3 (2026-09-03)

Principal components analysis now surfaces the most variable features it identifies. When PCA is run with the "Subset to top variable features" option, the variance-ranked list of features used for the analysis is saved and can be exported to a biomarker/target list directly from the PCA plot, either the full set or a top-n subset you specify.

Because features are ranked by variance across all samples in the experiment (independent of any single differential comparison), this gives you a way to capture the features that vary most across the whole dataset. The resulting list can then be used anywhere biomarker/target lists are supported (for example, as the gene source for a Clustering analysis heatmap).

No changes were made to the PCA calculation or plot data; this release adds the ability to persist and export the variable-feature ranking that PCA already computes.

Re-run an existing PCA to upgrade it to v1.3; prior results remain available as originally run. Exporting the variable-feature list requires running PCA with the "Subset to top variable features" option enabled.

v1.1 (2022-07-18)

Added the principal components summary measures table, which reports the proportion of variance explained by each principal component, as well as other measures. No changes were made to the calculation of PCA plot data.

v1.0

Principal components analysis (PCA) was implemented using the prcomp function in base R (v4.1). Users can toggle the Boolean values for scale and center. If selected by the user, CPM normalization is performed prior to running PCA using the cpm function in the edgeR R package (v3.36.0).

Sample correlation

v1.1 (2026-09-02)

Sample correlation results are now displayed as an interactive clustered heatmap, replacing the previous dot plot. Each cell is colored by the correlation coefficient between a pair of samples, and both axes are ordered by similarity and annotated with dendrograms, making sample groupings and outliers easier to see at a glance. When samples are organized into groups, that group metadata (e.g., Stage, Treatment) can be displayed as annotation bars alongside the matrix, and the heatmap color palette is customizable.

This release also resolves a data limitation in the previous version. A sample correlation result is a full pairwise matrix, and the earlier dot-plot format grew with the square of the sample count, so experiments with more than ~44 samples exceeded the plot's data limit and rendered with missing values. The heatmap is generated from a compact samples × samples matrix whose size grows only with the number of samples, so the complete correlation matrix now displays regardless of how many samples are included. The heatmap's color scale also adapts to the range of coefficients in your data, so typical high-correlation results are easier to distinguish than they were under the dot plot's fixed scale.

No changes were made to the correlation methodology (Pearson or Spearman, as selected) or to the underlying CPM normalization and feature filtering.

Re-run an existing sample correlation analysis to upgrade it to v1.1; prior results remain available as originally run.

Summary

v1.0

Summary analysis was implemented. No transformations to the data are run, the data is queried and displayed as-is in the user-uploaded assay data file.

Summary (CPM-normalized)

v1.0

Summary (CPM-normalized) analysis was implemented. The data in the user-uploaded assay data file is CPM-normalized using the cpm function in the edgeR R package (v3.36.0).