Skip to content

Latest commit

 

History

History
147 lines (123 loc) · 10.2 KB

nwbe_dataset_testing.md

File metadata and controls

147 lines (123 loc) · 10.2 KB
Raw

NWB Explorer testing with datasets

test datasets I'm using

Dataset ID Example files Title Example filename Notes
000004 example file single neurons recorded from the medial temporal lobes of 59 human subjects undergoing intracranial monitoring while they performed a recognition memory task. sub-P10HMH_ses-20060901_ecephys+image.nwb see below
000005 example file intracellular and extracellular electrophysiology recordings performed on mouse barrel cortex and ventral posterolateral nucleus (vpm) in whisker-based object locating task sub-anm184389_ses-20130207_behavior+ecephys.nwb see below
000006 example file mouse anterior lateral motor cortex (ALM) in delay response task sub-anm369962_ses-20170309.nwb see below
000007 example file A Cortico-cerebellar loop for motor planning sub-BAYLORCD12_ses-20180125T191601.nwb see below
000009 example file Maintenance of persistent activity in a frontal thalamocortical loop sub-anm184389_ses-20130207_behavior+ecephys.nwb see below
000011 example file Robust neuronal dynamics in premotor cortex during motor planning sub-255200_ses-20140910_behavior+ecephys+ogen.nwb see below
000016 example file Excitatory and inhibitory subnetworks are equally selective during decision-making and emerge simultaneously during learning sub-mouse1-fni16_ses-161220141515.nwb see below
000036 example file Allen Institute Openscope - Meaningful project
000039 example file Contrast tuning in mouse visual cortex with calcium imaging sub-661968859_ses-682746585_behavior+ophys.nwb see below
000048 example file Electrical and optical physiology in in vivo population scale two photon calcium imaging sub-222549_ecephys+ophys.nwb see below
000049 example file TFxSF tuning in mouse visual cortex with calcium imaging sub-661968859_ses-681698752_behavior+ophys.nwb see below
000067 example file Behavior-dependent short-term assembly dynamics in the medial prefrontal cortex sub-EE_ses-EE-042_ecephys.nwb see below
NWB Showcase Example example file NWBShowcase example file for 1D and 2D spatial series datatypes.nwb see below

Potential features/fixes

fixes

  • display experimenter ID and related publications in general tab
  • display contents of units and electrodes objects in details tab
  • display time values like file create date in general tab (need to fix display in date-time-metadata branch)
  • change color of comments in python console (fixed in osbv2 branch)
  • load files from dandi api links - issue #244

plotting enhancements

  • time series plots with multi-dimensional data - issue #63
  • single unit raster plots + mean waveform plots
  • behavioral events time series with markers+lines
  • imaging series - look into example datasets

other features

  • processing module tab
  • trials/intervals/epochs/units/electrodes table display
  • kill/quit command for plotting
  • load chunk of data to plot instead of whole session if big
  • see progress of download/file size/computation

related questions

  • what is the best way to add new plotting functionality?
  • what is the best widget to display table information?

Notes from test datasets

things to check for

  • all the general fields populate/all the content from the file is output somewhere
  • differences in plotting/display between explorer windows and nwbwidgets
  • what content is in the content index that could be added to acquisition/details/general

notes

dandiset 000004

  • content index info
    • electrodes/units display as objects in details tab
    • file create date not displayed
  • in nwb widgets
    • stimulus presentation shows up in widgets and series of pictures, errors out when try to visualize from content view
    • acquisition.events plots look different in the nwbwidgets vs the nwb plot output
    • acquisition.experiment_ids would want lines+markers to see data details since lots of similar values

dandiset 000005

  • content index info
    • show all content breaks the explorer, it just goes to a blank screen?
  • in nwb widgets
    • lick trace binary array does not show up in nwb window but does here
      • other binary arrays do get plotted normally so I'm not sure exactly what happened
    • related publications link works here, not showing up in general tab so not sure if it works there
    • single unit raster plot with nice spike times

dandiset 000006

  • content index info
    • intervals.trials.columns, electrodes.columns shows start times as list of [object, object]
    • spike_times_index, electrodes_index show up as VectorIndex type in content index list.
      • part of units.columns, but other rows of units.columns do not show up
      • not sure exactly what the VectorIndex vs VectorData means
  • in nwb widgets
    • related publications doi link does not work in general tab but does in the nwbwidgets
    • lick times shows up as straight line bc all values are 1, need lines+markers to visualize
    • units raster plots/tables, trial tables not visualized

dandiset 000007

  • content index info
    • spike times, obs intervals, and electrodes index for units columns dataframe
  • in nwb widgets
    • the location section of ogen_sites, electrode groups shows up as brackets/key value mappings in the nwb details tab
      • not sure if should be broken up into individual sections/columns or displayed like this
      • nwb widgets just shows the python console output which looks similar so might just be normal

dandiset 000009

  • similar to other files, tables of electrodes, units, intervals/trials don't display
  • think the main data in this file is from the single units

dandiset 000011

  • content index
    • example behavioral events time series where the data is all the same value but the time stamps indicate event times
    • seems like BehavioralEvents might be a data type that could be used as a condition to change the display to lines+markers
  • in nwb widgets
    • lots of good single units in this example and they display nicely
    • BehavioralEvents show up as single dots/markers for the relevant time stamps

dandiset 000016

  • lots of single trial broken up examples of roi response series in this one if that ends up being useful
  • in nwb widgets
    • the roi response series error out, but they work fine in the nwb explorer. Will have to check with HDFView to see if there is more than one value

dandiset 000039

  • content index info
    • multiple sections in details tab display [object, object], not sure if this will be fixed by the changes I made for the units column or if these are all separate issues
      • examples - DfOverF, intervals.epochs.columns
    • some modules(?) only in context index, ex. intervals, processing, imaging planes
    • device not displayed in general tab, not sure if it's very useful or helpful through
    • stimulus in the processing module here, not sure how common that is or if that's recommended
  • in nwbwidgets
    • some of the fields error out in the nwbwidgets so hard to say exactly what's going on with that

dandiset 000048

  • looks like it's just a placeholder dataset? for the ophys/ephys allen institute calibration data, not sure if it will be filled in the future

dandiset 000049

  • content index info
    • processing roi responses does not display all of them, only the last one
      • processing.brain_observatory_pipeline.data_interfaces.Fluorescence.roi_response_series.DfOverF
    • units.columns is a compositeList where the values end up being displayed as objects
      • similar thing with intervals.epochs except some of the columns are the vector index type
    • file_create_date is a composite list, could display as the time
  • in nwb widgets
    • imaging plane is a whole section here, but it does not show up in content index tab
    • devices does not show up in the general tab even though its filled

dandiset 000067

  • content index info
    • file_create_date - composite list, could make this more readable and put in the metadata
    • electrodes is a dynamic table type
      • vector data from electrodes.columns is displayed as a bunch of objects for each column
  • looked in NWB widgets
    • time values like 'timestamps_reference_time' and 'session_start_time' are not displayed anywhere

test dataset from showcase

  • spatial series only display first dimension even when 2D
  • some details have underscore instead of second space
  • all the units compared to nwb widgets look good