teese / eccpy Goto Github PK

Hey,
First of all, I really appreciate the tool, its quite useful, thanks for the effort :)
I find the usage of the tool quite confusing and the fact that I've to depend on both a gui-based tool (excel) and command prompt at the same time annoying, would it be a lot easier and logical if we could have a pure python version for it?
Something like a jupyter notebook where we could just use csv/pandas to do all the calculation and plotting?
This is more of a recommendation than an issue, nevertheless, I believe it would be a nice option to have.

The excel file holding the dose concentrations in the Versamax format cannot have any skipped rows that are "FALSE" for Contains_Data.

If any Contains_Data rows are False, followed by True, the wrong sample names are added!

Hi! Running your program on your test file generated_data_0.xlsx (Mac OS Big Sur, python 3.8.5) and get the following message: 'float' object has no attribute 'max'

Here are details:
AttributeError Traceback (most recent call last)
in
1 settings = "/Users/nomadkml/Desktop/result/ECCpy_settings_template.xlsx"
----> 2 eccpy.run_curvefit(settings)
3 eccpy.run_gatherer(settings)

~/opt/anaconda3/lib/python3.8/site-packages/eccpy/curvefit.py in run_curvefit(settings_excel_file)
77 print("Starting run_curvefit program for selected samples.\n")
78 for fn in dff.loc[dff["run curvefit"] == True].index:
---> 79 calc_EC50(fn, dff, settings, t20)
80 else:
81 print("None of the datafiles are marked TRUE for 'run curvefit'. Suggest checking the excel settings file.")

~/opt/anaconda3/lib/python3.8/site-packages/eccpy/curvefit.py in calc_EC50(fn, dff, settings, t20)
1107 # set the x-axis limit so that the legend does not hide too many data points
1108 # find the maximum dose concentration in the whole experiment for that day
-> 1109 maxAC = x_orig.max().max()
1110 # # obtain the variable altering the extension of the x-axis
1111 # x_axis_extension_after_dosemax_in_summ_plot = dff.loc[fn, "x-axis extension in summary fig_0"]

AttributeError: 'float' object has no attribute 'max'

Unclear why, but even with all dependencies installed the images of the curves appear, but none of the excel sheets showing the data appear. I should note the images have the EC50s.
May be a general excel issue as the template sheet only has the first data sheet read. The others are ignored. For instance, in the template, it reads dataset 0, but not 1 or 2.
Any suggestions greatly appreciated.

Problem: Running eccpy with the latest version of python-pandas gives the following error:
TypeError: read_excel() got an unexpected keyword argument sheetname

Eccpy should therefore be adapted to work with the latest version of pandas, numpy, and matplotlib.

Acceptance Criteria:

• eccpy tests work successfully with the latest anaconda python
  

• all other functions still work, and the plots still look okay.
  

Dear Mark Teese,

I have made 12 dose-response experiments, each following the scheme attached below.
I haven't noticed so far an option to merge n. replica in order to get mean & SEM curves in the current version of the script.
How should I prepare the ECCpy_settings_template.xlsx?

Problem: The compatible versions of pandas/numpy/matplotlib are not defined anywhere in eccpy. Breaking changes in the dependencies result in a non-functioning software.

Suggested implementation: after fixing Issue #10 , save a snapshot of the python dependencies and add them to a requirements.txt.

Acceptance criteria:

• pip and manual installation (python setup.py install) download the exact working dependencies

When replicates of a single sample are added (using the exact same sample name) the values for the response of a given dose are not averaged and the error bars (SEM) are not displayed in the graph produced.

need to check that max_acceptable_doseconc_stepsize_at_EC50 is taken seriously!

In one example, after collecting data from multiple experiments:

• the bar chart showed the wrong sample names on the x-axis, BUT
• the scatter plot with a point for each experiment was good
• the saved excel and csv were both correct

Need to check if the bug can be repeated, and find how the sample names became unordered.

The n_neighb is a key parameter for the automatic judgement of curves, but is really difficult to explain.

At some stage, it might make sense to use a value that is easy to explain in the settings file:
num_dp_expected_in_curve (number of datapoints expected in curve)
or even better,
num_dp_expected_at_inflection (number of datapoints expected at inflection)

If we don't want to change the code, n_neighb could simply be half this value:
n_neighb = num_dp_expected_at_inflection / 2

THe calculated value is correct, but not applied as a filter for TRUE/FALSE as good data.

At some stage, options like the method_calc_y50 should be choosable from drop-down menus.

It's relatively easy to do, and would greatly help users choose different settings

Hello teese,

Im trying to get the EC50 from some data on my own and I have not been able to do so with your program. The next error pops out everytime I run the code:index 0 is out of bounds for axis 0 with size 0
I sincerely do not know what I am doing wrong...even though i have read your tutorial over and over...
I would be very glad if you help me
Thanks in advance
pd: here is a screenschoot of the problem.