Is it possible to implement an auto-updater for the software? This would be useful for user without administration right on their university computers.

Installed all necessary libs on OS X with macports (wxpython-3.0)

ShapeOut works fine, but video files are not read.

Can be fixed by changing:
shapeout/gui/plot.py, line 290..292:

"
totframes = video.get(cv2.CAP_PROP_FRAME_COUNT)

video.set(cv2.CAP_PROP_POS_FRAMES, actual_sel-1)
"
The names of the constants changed from cv2.cv.CV_XXX to cv2.XXX in OpenCV 3.

By the way, to get ShapeOut running on OX X the following commands might help. Compilation takes some hours, but the libraries from macports run faster than those from anaconda. The problems with chaco and wx seem to be solved with the current version.

sudo port install python27 py27-ipython py27-scipy py27-matplotlib
sudo port install opencv +python27
sudo port install py27-wxpython-3.0 py27-statsmodels py27-kiwisolver py27-chaco py27-pip py27-simplejson py27-sip py27-macholib
sudo pip-2.7 install nptdms
sudo pip-2.7 install pyper

Then select python27 (macports) as standard python interpreter:

sudo port select --set python python27
sudo port select --set pip pip27

Write into ~/.bash_profile (Put in YOUR path to dclab)

!/bin/bash

export PYTHONPATH="${PYTHONPATH}:/Users/phi/git/dclab"

start ShapeOut with

pyhon shapeout/ShapeOut.py

This can be put into a .command file placed on the Desktop.

Sorting and grouping of plots on the panel would be helpful. E.g. by drag and drop or some kind of numbering.

There is an error appearing in AppVeyor:

rror: MD5 sums mismatch for download: http://repo.continuum.io/pkgs/free/win-32/mkl-11.3.1-0.tar.bz2 (6630258206482847d9e7af500a4c95cb != 52818e34fc742ca29649d88994258663)
Command exited with code 1

and last time:

rror: MD5 sums mismatch for download: http://repo.continuum.io/pkgs/free/win-32/freetype-2.5.5-vc9_0.tar.bz2 (f436ee7b29d97e4c7501eb084eb0ac92 != 2ba79f0e56db19388956a57463e8148c)
Command exited with code 1

Analyse where the program spends time.
Could it be that the KDE is taking a lot of time for larger datasets? - > Find a more performant way for estimating the densities (e.g. histogram 20 x 20 bins)

When importing and analyzing larger datasets, the ShapeOut gets quite slow.

Add number of events on each graph

just for convenience

Error-message:
...
line 349, in OnMenuExportPDF
AttributeError: 'PlotPanel' object has no attribute 'container'

Should be adjustable.
Please Paul :)

I want to plot data from a tdms file with fluorescence data. (Can't upload here, but can mail it.)
If I select X: Cell size, Y: FL-3 (red) Maximum and click "Apply", I get the following error message.

(...)
File "C:(...)build\win_shapeout\out00-PYZ.pyz\shapeout.util", line 52, in float2string_nsf
ValueError: Format specifier missing precision.

If the image shows tracked dirt or something, it should be possible to hide and exclude (not delete) the data point for analysis.

• Checks for updates on startup and every couple of hours
• Additional icon in the toolbar that pops up or gets enabled
• Checks for availability for current platform (Win/Mac/32bi/64bit) using GitHub api?

When adding measurements, add the option to filter for folder names. Useful if one folder contains a lot of measurements. Not sure if this can be done from within Windows.

Either display the cell image in the main window or find another way. Seeing the cells in the filter selection window makes it easier to determine the required filter polygon.

On smaller screens, the contour images are not visible.

Some measurements (recorded with the LabView setup) cannot be loaded with ShapeOut, because the Region tag in the *_para.ini files is missing (Region = "" default value?). No error is displayed to the user, because the directory is searched recursively in a separate thread. On a command line, this error pops up:

Traceback (most recent call last):
  File "/usr/lib/python2.7/threading.py", line 810, in __bootstrap_inner
    self.run()
  File "shapeout/gui/gaugeframe.py", line 82, in run
    res=result.get(timeout=.1)
  File "/usr/lib/python2.7/multiprocessing/pool.py", line 558, in get
    raise self._value
KeyError: 'Region'

A workaround is setting the region manually in the .ini file.

@phidahl @Naven2812 @DerDeef
Could you please fix this mistake made by the acquisition software? I believe it would not make sense to address this issue in ShapeOut.

1. I guess this is a bug? If I want to reset a polygon filter, nothing happens in the plot.
2. Also removing the polygon doesnt change something.

..this way one can find specific data points later

For each gated and plotted data set (e.g. with polygon filters), show statistical information:

• mean
• median
• mode
• std

... more @Naven2812 ?

Add checkbox to display results for local or global gating.

• add local per-plot gating (done in filter tab; commit 2c725e7)
• compare point clouds of different plots with several different analysis methods
  • linear mixed models (#15)
  • box plots (needs matplotlib and can be done externally, connected to #22)
• only display local gates when analysis/statistics tab is acive (too confusing)
• add tabbing to the plotting panel for display of graphs from analysis results

1. allow to use independent (polygon)-filters for each panel
2. allow independent statistics for each panel (choosable percentiles, mean, nr of points)
3. allow export of filtered data points in (two) column table for use in other plotting tools

After installation of 0.43 Shape-Out crashes during start
deinstallation does not help
0.42 also does not work anymore
0.40 works

in shapeout/plot.py line 292

video.set(cv2.CAP_PROP_POS_FRAMES, actual_sel-1)

The -1 has to be omitted for fRT-DC files, because frame numbering starts at 0. Is it possible to do this if there is any of the fluorescence channels detected in the tdms file?

fix recovering saved sessions

... to display individual populations of a data set in different plots.

When this option is checked, only the Contour accuracy of area is fixed. Contour accuracy of Def is not.

Allow to export individual plots as a PNG file.

There seems to be a chaco-related problem on at least Linux and Max OSx that makes such a feature currently unavailable:

enthought/chaco#295

It would help to be able to have an option which shows the original data (in faint grey or sth.) below the filtered data to see the impact of the filter settings.

• has to be packaged manually, because anaconda on mac does not like chaco
• use miniconda as base image
• find a tool that creates Mac app skeletons

one more thing.

Currently the aspect ratio is defined as the ratio of the cell elongation perpendicular to the flow by the elongation parallel to the flow. Could you invert this, so it is handled the same way as in the measurement software. Since elongation in flow direction is usually the one that grows, it somehow leaves you with a better intuition on the numbers and a somewhat broader scale. Example: one could go from 2 to 2.5 instead of 0.5 to 0.4 if more elongated (2.5x instead of 2x) cells shall be included.

This functionality will require a working R distribution and the R package "lme4".
The user needs to be able to enter 2 numbers (or even letters!) for each checked experiment to define the states "Treatment" and "TimeUnit".
The purpose is to get a number for the effect of the treatment and if this effect is significant.

I wrote the python function "LinMixMod" which takes a Model function, a list of response vectors (could be Area, Deformation...), a list of Treatments and a list of TimeUnits. The output is:
-Estimate: the average value of cells that had Treatment 1
-Fixed Effect: Change of the estimate value due to the Treatment 2
-Std Error for the Estimate and the Fixed Effect
-p-Value

Hi,

it would be great if we could display the fluorescence data on log axis

Cheers
K

Downsample data to a number of events given by user, e.g. all samples have the same number of data points

This means optionally fixing

• accuracy
• KDE parameters

Hi Paul,
the lateral position of the cells in the channel is currently connected to the position in flow direction. I think it would be best to change it to the position perpendicular to the flow. This way one can filter out those events that have not been properly centered. It would need a switch from channel "x" of the .tdms file to "y".

Thanks,
Christoph

As mentioned in #13 point 3, allow to export the filtered data set as tab separated values so that they can be plotted with other programs.

@chrherold
How complex should the exported data look like? Should the user be allowed to select multiple columns for export?

Easy to reproduce:

• open ShapeOut
• select a notebook tab other than "Filtering" and "Statistics"
• load data into the session
• select "Filtering" or "Statistics"

The notebook does not correctly draw its contents. Although the buttons etc. seem to exist (are just not visible).

In "Filtering", this bug is caused when the HT.HyperTreeList is added to the panel. As this seems to be a linux bug only, I will not attempt a fix for now.

It would be helpful if one can set different gates for different panels in one window for better comparison of data sets.

I believe the splash screen should be without transparency, because the image is so sparse.

I get this message when I start ShapOut build 1.0.194

Warning

iCCP: known incorrect sRGB profile

When hitting the reset-button in the analysis tab you get the error message:
...line 947, in GetDefaultConfiguration
KeyError: 'Analysis'

@phidahl Could you please provide me with a code snippet that extracts the required fluorescence information and point me to an example data set that contains such separate fluorescence curve tdms files?

Hi Paul,

we tried to analyse fRT-DC data with ShapeOut but as soon as we try to open the fluorescent parameter, ShapeOut starts working but gets stuck (though it doesn't display any error message). We would like to be able to load and dislay the fluorescence in a similar manner as deformation and size (especially with contour lines). Interestingly, there have been a few times when the analysis was working, but not consistently.

Many thanks in advance,
Maria & Ariane