Skip to content

v0.5.0 — open FlowJo workspaces, and choose the compensation matrix

Choose a tag to compare

View all tags
@david-priest david-priest released this 21 Aug 05:19
· 41 commits to master since this release
v0.5.0
7cd00d3
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Open a FlowJo workspace (beta)

Open Workspace… now accepts .wsp. A workspace holds gates and the names of the files they were drawn on but no data, so GateLab resolves those names against the files already open plus anything you point it at, and reports the tally. Samples whose FCS never turns up are named and skipped rather than blocking the others.

It asks about three things it cannot decide: which file is which (matched on the recorded path, then the sample node's name, then $FIL, because no one of those is present in every workspace and they frequently disagree), which strategy when a sample holds several, and which compensation.

Beta on purpose. Exercised against a 24-sample BD FACSDiscover S8 export and a published BD FACSDiva sort, both FlowJo 10.9 or 10.10 on macOS. Other vendors, older versions and Windows-authored workspaces are untested. Unreadable gate types are reported and skipped, never dropped silently.

The file's compensation matrix may not be the analysis's

This can change where your fluorescence gates land. A BD FACSDiva export writes the matrix recorded at acquisition into the FCS, while any adjustment the operator made afterwards lives only in the FlowJo workspace. In one published sort, seven of forty-two off-diagonal coefficients had been adjusted by hand, by as much as 0.105.

GateLab used to take the file's matrix without comment, which placed every fluorescence gate slightly wrong while looking healthy. When the two differ you are now shown both with the size of the disagreement and asked which to use, defaulting to the workspace's since that is what was in force when the gates were drawn.

Fixed

  • Gates on Comp- prefixed dimensions resolved onto the uncompensated channel and imported cleanly into the wrong space. Such a strategy is now either compensated properly or refused.
  • Gate outlines follow the real boundary. On an arcsinh axis spanning zero the drawn edge could miss it by hundreds of pixels, putting events on the visibly wrong side of a line that was gating them correctly. Only the drawing moved; gating, stored gates and exports are unchanged.
  • Contour plots no longer freeze while panning or stretching.
  • Reading a large workspace is no longer quadratic in its keyword count. A 24-sample workspace went from 25 seconds to 2.5.
Assets 2
Loading