ruairi-osul / drn-interactions Goto Github PK

Example

• Concurrent EEG/LFP spectrogram
• Concurrent EEG/LFP states
• Potentially add similarity plot

Coherance

• Quantify coherance between signals (potentially normalize)

Response overlap

• Show overlap in neuron response for spike rate and phase locking
• Adjusted mutual information for categories and spearman correlation for magnitude.

^{Created from VS Code using CodeStream}

^{Created from VS Code using CodeStream}

^{Created from VS Code using CodeStream}

^{Created from VS Code using CodeStream}

• Schematic figure
• Example raster
• Example PSTHs
• Response hists
• Consistency
• Clustering

• Pairwise
• Popcup
• Ensemble
• Graph Props

• Remove space for structural graph plots

• Duplicate
• Remove neuronal data
• Create graph figures
• Add graph figures
• Rearrange export

New Example

• Two example neurons: one phase locked and with not phase-locked. Take two neurons from the same recording (so you don’t need to show the EEG twice)
• Time series plot: EEG-raw, EEG-filtered, phase, neuron-1, neuron-2
• Phase description plot
• Neuron-1 phase histogram
• Neuron-2 phase histogram

Update Phase Histograms

• Troff to trough

• New notebooks without decoding
• Tidy up package code

How many neurons are in ensembles? Which Types of Neurons?

• Prop neurons in ensemble
• Prop by NT

Properties of ensembles

• NT Entropy
• Size
• Mean Weight
• Conductivity

• Single neuron variability

• Adjust current neuron types figure to have space for scatter
• Add large scatter showing spike rate, regularity and waveform width of each neuron.

^{Created from VS Code using CodeStream}

Pairwise

• Change in each comb
• Change in mag

Graph

• Graph metrics
• Node Metrics
• Edge metrics
• Graph classification performance v time shuffle

Ensembles

• Ensemble props
• Node classification in ensemble
• Edge classification in ensemble together

Example: State Variance

• Single neuron: histogram of spike counts by state
• Population: scatter plot of UMAP or (heat-maps, histograms, line-plots) of UMAPS in each state

State Decoding

• Add decoding from UMAPS or PCs
• (Decode without phase-locked neurons?)

State Dimensionality

• Intrinsic dimensionality of each state (paired bootstrap or by mouse).
• Do as a function of neuron size or bin width

Preferred phase decoding

• Transformer to bin phases into 4 bins around the oscillation
• Attempt to decode using spike data

^{Created from VS Code using CodeStream}

^{Created from VS Code using CodeStream}

^{Created from VS Code using CodeStream}

^{Created from VS Code using CodeStream}

Add neuron type volotility analysis and figure

Example Raster

• Raster at two or three points from a single recording

Example or demonstration of 30 mins activity

• Spike rate
• CV-ISI

Compare volitility by neuron type

• Spike rate
• CV-ISI

Spectral

• Spike rate of each neuron with frequency range 0.005 to 15
• Line plots or heatmaps per neuron type

Add schematic

^{Created from VS Code using CodeStream}

^{Created from VS Code using CodeStream}

Datasets

• Split datasets into first half and second half
• Perform check for both states in both halves

Single neuron example

• Phase histogram of single neuron first half second half

Population example

• Phase lock heat-map (SW only?)

Quantify

• Response adjusted MI
• Magnitude spearman
• Potentially by neuron type

• Example fig
• Response consistency

Add transformer that takes a df affinity matrix and reversibly returns a networkx graph

^{Created from VS Code using CodeStream}

^{Created from VS Code using CodeStream}

^{Created from VS Code using CodeStream}

Chose features

• Phase-lock theta
• Phase-lock delta
• Spike rate change

Cluster Plot

• Heat-map or scatter plot by cluster

Show distribution across neuron types

• Pairs bootstrap of MI or spearman correlation
• MI or spearman correlation by mouse

^{Created from VS Code using CodeStream}

^{Created from VS Code using CodeStream}

• change drn_interactions/interactions/graph.py

  • change density calculateion to normalize by number of edges, not number of nodes

• Recalculate BS graph

• Recalculate FS graph

• Upload fixed datasets

• (Add density to BS structure figure)

• (Add density to FS structure figure)

At the moment, if there are 0 spikes, then corr will return NA. This interfears with spectral clustering in ensemble detection

^{Created from VS Code using CodeStream}

Schematic

• Make schematic for drug assays

Rasters

• Add before and during rasters
• Mark neuron types

Heat-maps

• Sort by neuron type and add key (maybe bar plot?)

Histograms

• Add citalopram effect histograms
• Add WAY100365 effect histograms

Higher-order

• Potentially attempt to decode drug v pre drug in (CIT non-responders, SAL non-responders)

^{Created from VS Code using CodeStream}

• Duplicate
• Create Ensemble plots
• Add and rearange

^{Created from VS Code using CodeStream}