v0.9.6

[rafalkowalewski1]

Picasso 0.9.6 is here

Changelog

Important updates:

• Test subclustering plot (saved after G5M, can be plotted in Filter): fixed the labels of the plots
• Change of API in picasso.postprocess.nn_analysis: new inputs cause backward compatibility issues. The function now returns only the nearest neighbor distances, not the indices of the nearest neighbors.

Small improvements:

• Moved from merge sort to quick sort (usually faster due to lower memory usage)
• Render: increase the speed of picking circular locs, picking similar and filter by number of localizations (numba implementation)
• Render property histogram shown before rendering is activated
• Render property - removed legend
• Render Nearest Neighbor Analysis - saves nearest neighbors distances in the localizations .hdf5 file
• Render G5M dialog - adjusted the frame analysis checkbox
• Render G5M: removed the check for min. locs
• Render G5M: moved the check for too large clusters (or if any are present) before applying G5M to all channels (all channels analysis)
• Render masking: mask out saved area uses previously saved area if available in the metadata
• G5M documentation has been updated to include more troubleshooting tips and common issues, see here
• Localize zooms in and out centered at the current view
• Config file changes from 0.9.5 were documented and config template was updated
• SPINNA 3D masking: z slicing added for visual inspection
• SPINNA 3D homogeneous simulations automatically adjusts the observed density based on the z range set by the user and the xy area of the pick (if provided)
• SPINNA allows for different mask bin size and blur in lateral and axial dimensions
• SPINNA default mask blur of 500 nm in the API (previously 65 nm)
• Reduced copying and conversion of DataFrames to numpy arrays (less memory usage)
• picasso.io.load_locs and save_locs ensure that the saved metadata contains the required keys
• Updated documentation on filetypes and minimum requirements for HDF5 files and accompanying YAML metadata files in Picasso
• Use "col" in df.columns instead of hasattr(df, "col") to check for columns in DataFrames (better readability)
• picasso.postprocess functions picked_locs and pick_similar accept precomputed index blocks to speed up the picking of circular regions

Bug fixes:

• SPINNA 3D mask generation fixed (and picasso.render.render_hist3d)
• Test subcluster fix indexing
• Remove backward incompatible camera pixel size reading in SPINNA's mask generation (related to Request QoL feature: automatically read pixel size data if "Pixelsize" is defined in metadata even when user did not use Picasso for localization #602)
• Fixed localization masking for non-square mask (picasso.masking.mask_locs)
• Correct axial localization precision in Localize (magnification factor)
• Localize does not raise an error if QE is not found in the config file
• Localize does not automatically fit z coordinates if a 3D calibration file is loaded from the config file
• Render Test Clustering: fixed the full FOV button
• Fixed CLI picasso join

Notes

‼️ If installed in Program Files, Render and Localize may not be available for non-administrator users. Therefore, we recommend installing Picasso outside of Program Files. The current default location is C:\Picasso.
⚠️ Windows Safety features and Windows Defender may ask multiple times for permission during the installation and download.

---

This discussion was created from the release v0.9.6.