1.5.0 Release Notes

Highlights

  • Introducing a Normalization Platform – Now available in the Discovery band of the Workspace, which allows researchers to run normalization across parameter-sets in data matrices.
  • Multi-Geneset Enrichment Analysis – Simply select multiple genesets and run the enrichment platform against a geneset library (GMT) of your choice.
  • Opt-SNE (1) algorithm enabled! This means researchers won’t usually need to try and find optimal tSNE settings for their data if they use the default settings (Opt-SNE enabled). The algorithm uses KLD to find appropriate cut-offs in iterations.
  • Derived Observations of Genes used to identify genes in analytical genesets (gating in geneview) will now be included in Geneset inspector, and exports. This means that researchers can now export q-Value statistics, Fold-Changes, and more.

Improvements

  • Sorting available in Geneset Inspector.
  • Color Mapping now shown on same raw scale as parameter settings, regardless of what the maximum or minimum expression value in that parameter illustrates. This makes for more valid comparisons between plots showing the same color map parameter.
  • Support for bar delimited parameter names.
  • Large and small dot sizes now available in color maps.
  • Release notification mechanism to let users know when the next release becomes available.
  • Geneset Libraries (GMT files) can now be loaded into SeqGeq as a collection of genesets.
  • Parameter scaling default is now log. For non-derived parameters (gene and protein expression) this tends to work best.
  • Basic Tutorial documentation now included as a PDF along with demo data in the SeqGeq install.

Fixes

  • Functional KMeans cluster saving, and k number.
  • Saves should no longer lose connection to artifacts created during the analysis.
  • Color mapped backgrounds can now be set to “None” for a clear fill in plots.

Known Issues

  • Non-Unicode 8 characters (such as those in the Chinese alphabets) in a dataset’s file path will prevent data from being loaded into SeqGeq workspaces on Windows machines. Workaround: Make sure the file path leading to Gene Expression matrices are using only standard English character sets.
  • Heatmapping downsampled populations can lead to blank rows. Workaround: Downsample populations manually, as opposed to using the Downsampling platform when heatmapping downstream.
  • KMeans clustering is performed on the raw sample regardless of what subpopulation is selected. Workaround(s): Export the subpopulation of interest for computing KMeans clustering. Utilize an alternate clustering algorithm (e.g. FlowSOM).
  • KMeans clustering does not perform an autogate to illustrate clusters as populations in the workspace. Workaround: Run the AutoGateCategorical plugin on the KMeans clustering parameter to identify those subpopulations.
  • Gating path names longer than about 30 characters cause errors in calculating p-Values in Gene View Workaround: Name populations judiciously.
  • Legacy workspaces with gates drawn in Gene View will need to be re-confirmed, as changes in this version may have altered their position or geometry.
  • “Export Eigen Vectors” button from Principal Component type Dimensionality Reduction dialog for Windows will default to save the vectors in a temporary forlder. Workaround: Make sure to save vectors to a common location, such as Desktop.
  • Bring forward command in Layout Editor gives bug reporting window. Workaround: Close the bug report window and continue analysis.
  • Derived Observations (e.g. PCA vectors) calculated for certain Derived Parameters (e.g. a Gene Set in Cell View) will not be correctly applied when re-opening a saved analyses. Workaround: Don’t apply Derived Observations to Derived Parameters.
  • Only demo data can be loaded when running SeqGeq in demo mode; i.e. concatenated data (even from demo files) cannot be loaded into SeqGeq in demo mode.
  • Some hotkeys meant to open tabs in the Workspace will generate a bug report window.
  • Bug reporting window does not send information to FlowJo, LLC. Workaround: Send an email to tech support for a speedy reply and troubleshooting: seqgeq@flowjo.com
  • The 24 hour authentication grace period for temporary offline access will sometimes fail to correctly activate.
  • Running tSNE on a raw expression matrix is not advised. Workaround: We recommend running tSNE on principal components, after running PCA on the expression matrix of interest.
  • On a Mac, R-based plugins fail to calculate if the file paths contains a space or other special characters. Workaround: Remove spaces and special characters from file paths associated with the analysis.
  • Parameter Selector for plugins won’t discriminate between identical samples opened in different workspaces (and uses the first workspace only). Workaround: Don’t open the same dataset in different workspaces simultaneously while attempting to access plugins.

References

1) Belkina, Anna C., et al. “Automated optimal parameters for T-distributed stochastic neighbor embedding improve visualization and allow analysis of large datasets.” bioRxiv(2018)