oscarbranson / latools_gui Goto Github PK

Tweak the behaviour of pane resizing when the window is resized.

Ideal Behaviour:

• navigation Pane has fixed height and scales horizontally with window.
• controlsPane has fixed height and scales horizontally with window.
• graphPane scales vertically an horizontally with window.
• options (samples list and analyte list) are pinned to L and R sides, and graph itself expands to fill extra space in middle.

Control panel for statistic calculation and data export.

Features:

• Select statistics to be calculated using check boxes.
• Export 'raw traces' from any stage of analysis for external analysis/plotting.
• Export 'minimal dataset' as a .zip file.

By default each stage of analysis is applied to the data in the focus attribute. In some stages of analysis, the user may want to apply the operation to a different stage of analysis.

For example, if the user wants to re-do 'despiking', it makes more sense to apply the second round of despiking to the raw data, rather than the previously despiked data.

Implementation

To allow this, we should add an 'apply to' drop-down menu, which contains all the currently available stages of processed data.

Suggest adding this on the lower right of the controlsPane, just to the left of the 'apply' button?

Should be available in the following stages:

• despiking (#29)
• autorange (#31)
• bkg_calculation (#32)

After background subtraction, this option is no longer appropriate, as later stages of analysis have to be based on the background corrected data.

On Selection

The stage of analysis should be passed as a focus_stage variable to each of the appropriate functions.

Prerequisites

Some of these functions don't currently accept a focus_stage argument. Follow oscarbranson/latools#19 for progress on this issue.

Everything has to be documented!!

Both on the user-facing side, and the back end (so I can maintain the software!)

Ideas:

• Use sphinx - already set up in repo.
• Can existing function docstrings and argument descriptions be made available through GUI?

Pop up to visualise calculated background.

Interacts with latools.analyse.

• Able to select analytes.

File

• save
• load
• exit

Configuration

• make

Control panel for Despiking.

Needs to be able to handle all arguments of latools.despike

Create blank templates for all elements of the GUI, which can be populated with controls for each stage of analysis.

Templates needed for:

• Data import (#8)
• Analytical control panel.
• Plot pop-up (for #7, #11, #14).
• Undockable plotting window (for #4).
• Filtering panel (#16)

All templates must be documented and well-commented.

• Explore PyQt style sheets.
• Make style sheet for GUI (requires progress on #1)

Create a functioning plotting PyQtgraph panel that replicates the functionality of latools.helpers.plot.tplot. Panel will be displayed on lower half of GUI through all stages of analysis. Panel should be able to 'undock' from main window, to facilitate multiple displays.

Needs to interact with latools.analyse (analysis-level) and latools.D (data container) objects.

Should grab data directly from the latools.D data object (all references below are to attributes of a D instance).

Should grab analyte info and list of samples and standards from latools.analyse object.

• Draft plot that displays data from .data['focus_stage'].
• Ability to switch between samples (scroll list on left hand side?)
• Analyte on/of selection checkboxes (right hand side?). Must automatically populate from .analytes, and assign unique colours.
• Controls for various options of tplot

Autodoc must be configured within Sphinx to read docstrings from the project and format them into neat documentation.

Autodocs will still require some continuous, manual effort to add pointers to each module/class/function, and to organise them neatly within the Sphinx documents.

Control panel for calibration.

Needs to handle all arguments to analyse.calibrate.

• Populate available SRMs from SRM databse. Choose with check-boxes.
• Need way to visualise calibration (see #14).

Unsure how feasible this is with GUIs, but if possible:

• Build test-suite alongside
• Set up Continuous Integration / testing (e.g. with TravisCI) to check all new commits on Linux/Mac/Windows systems.

Docstrings in the numpy format need to be included for every module, class, and function that we have already created. In future these docstrings should be created as you work, so that we aren't left with a big pile of documentation work near the end.

Numpy format: https://github.com/numpy/numpy/blob/master/doc/HOWTO_DOCUMENT.rst.txt#docstring-standard

Filtering control panel.

Must be able to:

• Create filters in any order.
• Individually activate/deactivate filters for different samples/analytes.
• Visualise effect of filters.

Control panel to choose denominator analyte and calculate ratios.

Needs to deal with all arguments to latools.ratio

Control panel for Autorange function.

Needs to handle all arguments of latools.autorange

Data displayed in the plot should update as processing is applied.

Expected Behaviour

• Data in the plot should update whenever 'Apply' is clicked in the controlPanel
• When changing the focus of the controlPanel stack, the data in the plot should also change if that stage of analysis has been applied.

For this second one, I've added an attribute to the analyse class to record which stages of analysis have been conducted (analyse.stages_complete). So perhaps something like:

<click> NextPanel(name_of_next_stage):
  if name_of_next_stage in analyse.stages_complete:
    update_plot()

You can access function defaults using inspect (in the standard library)

For example:

import inspect
import latools as la

# get function or method signature
sig = inspect.signature(la.analyse.despike)
# collect default arguments into dict
{
    k: v.default
    for k, v in signature.parameters.items()
    if v.default is not inspect.Parameter.empty
}

>> {'autorange_kwargs': {},
    'expdecay_despiker': False,
    'exponent': None,
    'exponentplot': False,
    'maxiter': 4,
    'nlim': 12.0,
    'noise_despiker': True,
    'win': 3}

Can then use this dict to populate defaults?

Have the graph instantly update when settings are changed (current sample, visible elements, etc)

Two sub-panes ('boxes?'), one for each depsiker.

Check box at top of pane, with options underneath

Box 1:

• exponential decay despike - check box
• exponent - number (optional, default blank)
• maxiter - integer (default = 4)

Box 2:

• noise despike - check box
• 'win' - number (default = 3)
• nlim - integer (default = 12)
• maxiter - integer (default = 4)

Pass arguments to method of analyse instance:

despike(self, expdecay_despiker=False, exponent=None,
        noise_despiker=True, win=3, nlim=12., exponentplot=False,
        maxiter=4, autorange_kwargs={}):

Oscar: Need to implement separate maxiter for each despiker

• New Project (define name)

• Open Project

• Recent projects dropdown

• Link to online manual

Create labels for plot items, or make it so the user can hover over them to display info.

Get them to appear again

[Message from Margaret - if Ilya and Zane could create perhaps a check list of things they have done this week for this issue, that would be good! this just lets us and our many eyes know what we've been doing!]

• example one finished
• example two not finished
• ...

• Import

• De-spiking

• Autorange

• Background

• Ratio

• Calibration

• Filtering

• button to open the database file location

• Graph out full workflow and user interaction.
• Identify specific elements/panels which need making.
• Communicate needs to help #3.

• analyte - drop down (including all analytes in data, and 'total_counts' option, default=total_counts)
• gwin - integer (default = 5)
• swin - integer (default = 3)
• win - ineger (default = 20)
• on_mult - double number boxes (default = [1., 1.5])
• off_mult - double number boxes (default = [1.5, 1])
• nbin - integer (defualt = 10)
• log tranform - check box (default = checked)

Arguments passed to method of analyse instance:

autorange(self, analyte='total_counts', gwin=5, swin=3, win=20,
          on_mult=[1., 1.5], off_mult=[1.5, 1], nbin=10,
          transform='log', ploterrs=True)

ploterrs

Need to have some way of handling autorange failures. Let's talk about this later

Make it so that the created external graph live updates with the main graph.

• Swap current sample

• Switch current stage

• Update function changes the graph's state based off arguments

UI buttons are able to interact and update the Graph object

• Apply buttons

• names/desc done and put into gui for all stages

• Import data

• Despike data

• Autorange

• Background Calculation / Subtraction

• Ratio

• Calibrate

• Main plot panel controls

• Background plot controls

• Calibration plot controls

• put them in json file so they can be easily updated

There should be three package types for final distribution

• The standard LAtools package, run from the command line by advanced users like Oscar.
• The intermediate package, which runs the LAtools GUI through the Python interpreter (probably launched by a shortcut or bash script). Assumes the user has installed Python and will be able to run a pip command to install all dependencies.
• The basic package, which is a stand-alone folder containing the PyInstaller compiled package of the LAtools GUI. Assumes no dependencies or skills on the behalf of the user.

Options that need to be implemented in the Data Import Stage:

• Browse needs to open file browser, and be able to select a directory.
• Config - drop down menu with options populated from latools.cfg.
• SRM Identifier - text field.
• file extension - text field
• Internal Standard - text field.

Arguments passed to the initialisation of latools.analyse:

__init__(self, data_folder, errorhunt=False, config='DEFAULT',
             dataformat=None, extension='.csv', srm_identifier='STD',
             cmap=None, time_format=None, internal_standard='Ca43',
             names='file_names', srm_file=None)

Specify internal_standard during ratio stage (#35), not during import.

Control panel for background calculation and correction.

Needs to offer both latools.bkg_calc_weightedmean and latools.bkg_calc_interp1d, and offer different arguments depending on choice.

• Way to switch modes within panel, and display different arguments (talk to #3)
• Need a way to plot calculated background (see #11)

Parameters:

• drift_correct - check box (default = checked)
• srms_used - scroll list with check-boxes, populated from srm database file.
• zero_intercept - check box (default = checked)
• n_min - integer (default = 10)

Extra:

• option in menu bar to reload srm database?

Pass parameters to method of analyse instance

calibrate(self, analytes=None, drift_correct=True,
          srms_used=['NIST610', 'NIST612', 'NIST614'],
          zero_intercept=True, n_min=10)

• data Load
• despiking
• autorange
• background correctoin
• ratio
• calibration
• filtering?
• export

Also:

Think about 'apply to' option for early analysis stages (e.g. despiking)

Create legend in graph widget that displays elements and their colours.
Have scroll bar if elements goes beyond the window

Panel to display cross-plots.

Should be pop-up, as won't fit well in current layout (requires square window, which would displace everything).

Needs to work at both analysis and single-sample level. I can help work this out.

• Create panel (talk to #3)
• Display crossplots at whole-analysis and sample level (be able to switch between)

• internal_standard - drop down, populated from available analytes.

Pass arguments to method of analyse instance:

ratio(self, internal_standard=None, focus='bkgsub')

Pop-up displaying calibration plot.

Similar to existing calibration_plot.

Combine frameworks developed by Simon and Zane & Ilya into a comprehensive framework.

Background Correction

There are 2 stages: bkg_calc and bkg_subtract

Background Calculation

• method : drop-down containing ['bkg_calc_weightedmean', 'bkg_calc_interp1d']

options populate depending on which method is chosen.

bkg_calc_interp1d:

• kind - integer (default = 1)
• n_min - integer (default = 10)
• n_max - integer (optional, default = blank)

bkg_calc_weightedmean:

• weight_fwhm - float (optional, default = blank)
• n_min - integer (default = 20)
• n_max - integer (optional, default = blank)

Universal options (for both):

• cstep - float (optional, defauly = blank)
• bkg_filter - check box (default = unchecked)
  If bkg_filter is checked, two more options:
• f_win - integer (default = 7)
• f_n_lim - integer (default = 3)

After all options are completed:

• Button saying 'calculate background'
• Button to plot background in popup.

Arguments passed to one of two methods in the analyse instance:

bkg_calc_interp1d(self, analytes=None, kind=1, n_min=10, n_max=None, cstep=None,
                      bkg_filter=False, f_win=7, f_n_lim=3)

bkg_calc_weightedmean(self, analytes=None, weight_fwhm=None,
                          n_min=20, n_max=None, cstep=None,
                          bkg_filter=False, f_win=7, f_n_lim=3)

Background Subtraction

• button to subtract background

Arguments passed to method of analyse instance:

bkg_subtract(self, analytes=None, errtype='stderr', focus='despiked')

Data import interface at splash screen.

Need to be able to specify all options in the __init__ of latools.analyse.

Needs to be able to handle:

• Folder of csv files.
• 'Reproduce' export (currently a file and folder, moving to a zip file soon).