Skip to content

Release notes

The authoritative, complete list of changes is CHANGELOG.md in the repository. This page summarises what a user of PaintOmics 4 will notice that is different.

Unreleased — PaintOmics AI

The successor to PaintOmics 4 has not been cut as a numbered release yet. Everything below is on master. Which of it is switched on depends on the server you are using — see What the AI does.

New analyses

  • AI interpretation. An agent walks the network of every KEGG, Reactome and OmniPath interaction for your organism with your values on its nodes, and a writer turns the walk into a Results section with numbered citations. Every value, leg and citation is checked by code before you see it. See The interpretation.
  • Metabolite class activity. Whether a whole KEGG BRITE class responded, at three levels of the hierarchy — a permutation test on your own replicates when the upload has them, a binomial test on your relevant list when it does not. See Metabolite class activity.
  • Metabolite hub analysis. Which metabolites have differentially expressed genes concentrated around them in the KEGG reaction network. See Metabolite hub analysis.
  • Regulatory modelling with MORE, behind a method chooser with four engines — PLS1 and MLR, each on the R reference implementation or on a Rust port — plus a regulator–target network on the results screen and regulation-per-condition tables that hand off into the pathway view. See Regulatory omics.
  • OmniPath as a fourth pathway database. See The OmniPath interaction network.

New in the workflow

  • Any number of conditions. Analyses run across as many conditions as your files carry, not two, with per-condition significance stars in the heatmaps, in the hub analysis and in class activity, and a weights panel for the combined p-value.
  • Replicates. Per-sample values with an experimental design, aggregated for the visualisation and used directly by MORE and by the class activity test.
  • An input-format check on every file you pick, with one-click deterministic repairs for mechanical faults, and — on servers that have turned it on — an AI converter for files that are not in PaintOmics' format at all.
  • Compound disambiguation on Step 2, with an AI suggestion for the ambiguous names in one pass (capped at 90 sets per run, so a job with hundreds of them takes more than one).
  • Example datasets for every pipeline, generated from a manifest that also records the pathways the analysis should recover. See The example datasets.
  • Jobs are kept for as long as the interface says they will be — 7 days for a guest job, 14 for one belonging to an account — and no longer.

New in the interface

  • The application is renamed PaintOmics AI. Navigation moved into the header and the left rail was removed.
  • A dark theme, with a toggle in the header.
  • A contents sidebar on the results screen that follows the database tab you are looking at.
  • An organism picker that ranks matches by what you typed.
  • A rebuilt heatmap colour scale, fitted to the range your data actually occupies, and the same ramp now paints the pathway diagrams.
  • A progress bar for a running job that only moves forward and says what the job is doing.
  • Browse… moved inside the file field, and every file row now says whether the job requires that file.

Under the hood

Python 3.11, Flask 3 and pymongo 4 against MongoDB 7; a Docker Compose deployment with nginx and TLS; and continuous integration with a pull-request gate and a nightly regression run against recorded baselines.

PaintOmics 4

  • New tutorial video — 18 July 2022
  • New user guide — 1 July 2022
  • Release of PaintOmics 4 (v1.0.0) — 31 May 2022