Gnomesh Product Roadmap

Highlights

• A real plot editor. Double-click a Plot node — or use Open plot editor… — and the plot opens full-screen with room to work. A searchable list of sections down the left, that section's controls top right, and whatever you clicked on the plot below them. Both side rails collapse when you want the figure to have the screen. The node itself is now a clean summary with a thumbnail of the current plot.
• Everything redraws as you type. Titles, colours, point sizes, axis ranges, padding — the plot follows the keystroke instead of waiting for you to click away. Choosing a different column re-runs the node for you when its data is already loaded, and there is a Run button in the editor header for when it isn't, so you never have to close the editor to see a change.
• Several series on one plot. + Series plots another column against the same X axis, each series with its own name and colour, and the legend and title keep up as you add them.
• Figure presets. Start from Screen, Paper — single column (85 mm at 300 dpi, 8 pt type), Paper — double column (174 mm at 300 dpi) or Presentation (1920 × 1080, large type for a projected slide). Save your current settings as a preset of your own, star one as your default so every new plot begins there, and export or import presets as JSON. A preset only changes what it names, so a paper style applies to any kind of plot without replacing the plot type. Presets are stored with your account, so they follow you to any browser you sign in from.

Working on the plot itself

• Drag your labels. Ctrl-drag any label — the title, an axis label, an annotation — to exactly where you want it, and it stays there through re-runs and reloads. A reset control puts it back.
• Drag a label to where it belongs. + Data label and + Note are drag sources: drag one onto the plot and a marker follows the pointer, with a data label snapping to the nearest point and naming its values, so you can see what it will attach to before you let go. Clicking the button still drops one in the middle as before.
• A key to the canvas gestures. The editor header carries a help icon listing every gesture the plot understands — pan, zoom, move, select, edit in place — so none of them has to be discovered by accident.
• Click to edit what you clicked. Selecting an element on the canvas opens its own controls — text, font size, colour, position, angle, leader line — in the panel beneath the section controls.
• Zoom one axis at a time. Scrolling over the plot zooms both axes; scrolling over an axis gutter scales just that axis.
• Legend inside or outside. Place the legend over the plot or beside it, where the plot area makes room for it, and click an entry to hide or show that series.
• Hover for values. Pointing at a point reads out its values, including any extra fields the data carries.
• Axis labels write themselves. Titles and axis labels default from the columns you bound — Intensity vs Retention Time — and a label you have written yourself is never overwritten by a later change.
• Export at the size you meant. PNG, JPEG and SVG, with the export dimensions set in the editor and honoured on the way out.

Fixes

• Cosmetic edits no longer discard your plot. Changing a title or a colour after a run re-renders in place; only changing which columns are plotted asks for the data again.
• Plot settings stay saved. Values you had entered are no longer replaced with defaults when the panel refreshes, and a second edit lands on the same plot as the first.
• Adding a node no longer breaks the next run. Adding or rewiring a node and pressing Execute used to fail with a "node not found" error until the page was refreshed. Rewiring an already-run node also gave a result computed from the old connections — quietly, with no error. Both are fixed.
• Moving a label to a corner keeps it on screen. Setting a label’s position to a corner used to hang most of the text off the edge of the plot. A corner now means the label’s matching corner, so it lands where you asked.
• Dragged labels render cleanly. A label moved to certain positions could come out with broken strokes and a stem stepping sideways part-way down, most visibly on displays at 125% scaling. Text now lands on the pixel grid wherever it is placed — which also sharpens the rotated y-axis title in every plot, moved or not.

Highlights

• Categorical bar charts. A text column — sample names, conditions, genes — can now be the X axis of a bar chart, with the bars labeled by your actual category names. Columns that can't be plotted numerically are called out clearly, down to the exact cell.
• Run Node handles everything. The inspector's run button executes any not-yet-run upstream nodes automatically and works from a fresh workflow — no need to run anything else first. The button reads Run Node or Re-run Node to match whether the node has produced output.
• Pop-out inspector. The inspector can detach into its own window (Chrome/Edge) and keeps following your node selection live — handy on a second monitor while you work the canvas.

Workflow Editor

• A better-behaved inspector. The floating panel sizes itself to its content, its header controls are always reachable, and every section reads correctly in light and dark themes.
• Tooltips everywhere they belong. Tooltips are never cut off by a panel edge and reposition themselves when space is tight.
• Missing values are handled per row. A gap in your data skips just that row of the plot, and the warning states exactly how many rows were left out of which column.

Files and Notifications

• Smarter duplicate handling. Re-adding a file the project already holds just tells you it's there — the overwrite question only appears when the content actually differs.
• Notifications wait for you. Hovering a notification holds it open until you move away, and every notification can be dismissed.

Accounts and Admin

• Live password feedback. Registration shows the password requirements as a live checklist with a strength meter, updated as you type.
• Clearer server-error visibility for admins. Each entry in the error log expands to show when it happened, to whom, and whether the alert notification went out — plus a test control to verify the alert pipeline end to end. Alert delivery itself is more dependable.

Highlights

• Much less waiting on load. Everything the site sends is now compressed — the main application binary arrives in under half the bytes it used to — files you already have are re-used between page views instead of downloaded again, and the footer no longer pulls in the entire analysis runtime just to print a version number.
• The workflow canvas tells you how to start. An empty canvas now says what to do, and the insert menu explains what each node is for and when you would reach for it — with arrow-key navigation and a keyboard shortcut (Ctrl+Space) to open it.
• Protein coverage maps. A Digest node can now be plotted directly. Connect Text → Digest → Plot and you get a coverage map of the peptides across the protein sequence, grouped by missed-cleavage count.

Workflow Editor

• Empty-canvas hint. Shown only while the canvas is genuinely empty, and it waits for a saved workflow to finish loading rather than flashing on the way in.
• Node help in the insert menu. Every node carries a short "when would I reach for this" description alongside its inputs and outputs. The panel remembers whether you left it open.
• The advanced view is now the default. New accounts land on the full workflow canvas instead of the simplified view.
• Faster plot framing. Reset view, auto-size and range framing no longer re-scan the data in the browser — the plot answers from what it already knows, which is most noticeable on large selections.

Accounts and Admin

• Fixed the defaults a brand-new account starts on — theme and page size could come up empty or zero before the first save.

Fixes

• When a part of the application fails to load, you now get a message saying so instead of a page that silently does nothing.

Highlights

• Large vendor files just open now. Thermo RAW acquisitions that used to stall or run out of memory — including multi-gigabyte runs with tens of thousands of scans — load and plot. Spectra are read as you look at them instead of all at once, so opening a file is quick and stays responsive as you scroll.
• Faster, smoother plots. Plotting is GPU-accelerated end to end. Zoom, pan, cursor readout and click-drag range selection all happen on the plot itself, so large chromatograms and spectra stay fluid.
• In-silico digestion and LC-MS/MS simulation. Two new workflow nodes: Digest takes a protein sequence (raw residues or FASTA), applies the cleavage agent and modifications you choose, and produces peptides. Simulation turns those peptides into a synthetic MS run — charge states, fragment ions, isotope envelopes, gradient elution, an MS1 scan grid and DDA MS2 scans. Useful for testing a workflow before you have instrument data, or for sanity-checking what you expect to see.

Mass Spec Explorer

• Peak labelling with a sensitivity slider. Label peaks on the TIC and on the spectrum, and dial detection up or down without leaving the plot.
• Auto threshold. One button picks a sensible signal-to-noise cutoff for the current data and re-frames the view around what it found.
• Reset view and auto-size buttons for getting back to the full range, or framing the detected peaks.
• Fast display toggle. Wide selections are approximated for speed by default; turn it off in the info bar when you need exact points, and the choice sticks.
• Performance improvements. Scan tables, spectrum extraction and export were reworked, and are considerably faster on large files.

Workflow Editor

• New nodes: Digest, Simulation, Comparison, Text and Export, alongside the existing File, Parse, Mass Spec, Plot and Debug nodes.
• Comparison got smarter. Text comparisons now use a proper line diff rather than a rough character-overlap score, with an ignore whitespace option. Mass spec comparisons gained a 2D retention-time × m/z heatmap view, with retention-time bin width and drift tolerance you can tune.
• Run timer. Workflows show elapsed time while they run.
• Full-output download. The Debug node's file export now writes the complete output — every column, every scan, every line — with progress shown on the button, instead of a truncated preview.

Accounts, Projects and Site

• Terms of Service, Privacy Policy and Refund Policy are now published on the site, and downloadable as PDFs.
• Friendlier 404, 429 and 500 pages.
• Notification preferences, the project header and the settings page were reworked; login, registration and password-reset flows are more consistent and give better errors.
• The app version is shown in the footer.

Fixes

• Fixed workers becoming unresponsive after an internal error — this is now detected and recovered automatically, instead of leaving a workflow stuck.
• Fixed the m/z plot coming up blank for some RAW files.
• Fixed spectrum failures being silent; they now surface as a visible notification, and the session keeps working.
• Fixed scans with an unspecified MS level being excluded from MS1 queries.
• Fixed the Data Explorer freezing on deeply nested data.
• Changing your password no longer logs out your active session.
• Fixed notification preferences reverting immediately after being changed, and notifications showing regardless of those preferences.
• Numerous stability and memory fixes around parsing, plotting and file handling.

Notes

• The older Datasets view has been retired. The workflow canvas is the place to load and inspect data; everything the Datasets view did is available there.
• workflow.json is reserved — project uploads using that filename are rejected.

• Node-based workflows. Build an analysis by wiring nodes together on a canvas, and run it without leaving the page.
• Mass spec analysis. Load instrument files, browse scans, and inspect chromatograms and spectra.
• Custom plot generation from any workflow output.
• Projects and accounts. Keep files and workflows organised per project, with registration, login and password recovery.