john-judge / TSM-to-ZDA

Reads in TurboSM data files (.tsm, .tbn), bins to an image size that the original C++-only PhotoZ can handle, translates or fills in metadata, and writes a legacy ZDA file

MIT License

0 stars 0 forks source link

Signal comparison: kv2.1hVOS3.0/PV-Cre versus Ai35hVOS/PV-Cre #8

Open john-judge opened 1 year ago

john-judge commented 1 year ago

kv2.1hVOS3.0/PV-Cre

8/19/22

9/1/22

Date	Slice/Loc	Mean snr	Number ROIs	Line
09/01/22	05_01	3.041	25	hVOS 3.0 / PV-Cre
	06_01	3.349	30	hVOS 3.0 / PV-Cre
	07_01	3.280	15	hVOS 3.0 / PV-Cre

ai35hVOS/PV-Cre

8/22/22

6/21/22

Slice/Loc	Mean snr	Number ROIs	Line
01_01	4.157	45	ai35hVOS /PV-Cre
02_01	4.901	35	ai35hVOS /PV-Cre
03_01	5.356	30	ai35hVOS /PV-Cre

Slice 1 Loc 1

	Pixel	Occurrences
5 73	4.0	44.44%
16 52	4.0	44.44%
22 79	4.0	44.44%
15 39	4.0	44.44%
16 53	4.0	44.44%
13 62	4.0	44.44%
4 54	5.0	55.55%
15 52	5.0	55.55%
12 70	5.0	55.55%
16 60	6.0	66.66%

Slice 2 Loc 1

	Pixel	Occurrences
14 67	4.0	57.14%
17 61	4.0	57.14%
0 64	4.0	57.14%
12 59	4.0	57.14%
13 65	4.0	57.14%
2 66	4.0	57.14%
16 58	4.0	57.14%
15 67	4.0	57.14%
14 70	4.0	57.14%
14 66	6.0	85.71%

Slice 3 Loc 1

	Pixel	Occurrences
51 74	3.0	50.0%
51 75	3.0	50.0%
34 68	3.0	50.0%
46 74	3.0	50.0%
29 16	3.0	50.0%
37 22	3.0	50.0%
41 70	3.0	50.0%
34 69	3.0	50.0%
21 26	3.0	50.0%
6 11	4.0	66.66%

john-judge commented 1 year ago

To do:

[x] Repeat this comparison with SNR.
[ ] Set up trace discarding and near-electrode exclusion

And later:

[ ] kv2.1 comparison by looking at HC slices between kv2.1hVOS1.5 and ai35hVOS. Consult Wen-Chi
[ ] ~Set up better df/f method that bypasses PhotoZ RLI division. Take a dark frame manually, use to estimate the resting light constant.~

SNR Comparison: Results

Date	Slice/Loc	Mean SNR	Number ROIs	Line
08-19-22	01_01	3.47	60	kv2.1hVOS3.0/PV-Cre
	02_01	3.20	30	kv2.1hVOS3.0/PV-Cre
09-01-22	04_01	2.17	30	kv2.1hVOS3.0/PV-Cre
	05_01	4.22	25	kv2.1hVOS3.0/PV-Cre
	06_01	2.85	30	kv2.1hVOS3.0/PV-Cre
	07_01	2.63	15	kv2.1hVOS3.0/PV-Cre
08-22-22	01_02	3.33	25	ai34hVOS/PV-Cre
	02_01	2.91	10	ai35hVOS/PV-Cre
	03_01	3.17	15	ai35hVOS/PV-Cre
	04_02	2.35	15	ai35hVOS/PV-Cre
06-21-22	01_01	4.16	45	ai35hVOS/PV-Cre
	02_01	4.90	35	ai35hVOS/PV-Cre
	03_01	5.36	25	ai35hVOS/PV-Cre

Summary

Line	Mean SNR by ROI	Total ROIs	Slice-to-slice SD	Notes
kv2.1hVOS3.0/PV-Cre	3.31	160	0.16
ai35hVOS/PV-Cre	4.05	170	1.10	Includes 6/21/22
ai35hVOS/PV-Cre	3.00	65	0.430	Excludes 6/21/22

Update

Justification to exclude 6/21/22 data found. Lab note entry on 7/22/22: "set up 200 us stim iso" This means that all new rig experiments 7/19/22 and early used 1000 us duration, all after are use 200 us.

john-judge commented 1 year ago

10/21/22 SNR

Date	Slice/Loc	Mean SNR	Number ROIs	Line
10-21-22	01_01	3.222	30	kv2.1hVOS3.0/PV-Cre
	01_02	3.464	30	kv2.1hVOS3.0/PV-Cre
	01_03	3.456	15	kv2.1hVOS3.0/PV-Cre

10/28/22 SNR

Date	Slice/Loc	Mean SNR	Number ROIs	Line
10-28-22	01_01	3.500	80	kv2.1hVOS3.0/PV-Cre
	02_01	2.059	71	kv2.1hVOS3.0/PV-Cre
	03_01	1.661	50	kv2.1hVOS3.0/PV-Cre

11/04/22 SNR

Date	Slice/Loc	Mean SNR	Number ROIs	Line
11-04-22	01_01	3.037	100	kv2.1hVOS3.0/PV-Cre
	03_01	1.573	95	kv2.1hVOS3.0/PV-Cre
	04_01	1.742	100	kv2.1hVOS3.0/PV-Cre

11/09/22 SNR

download

Date	Slice/Loc	Mean SNR	Number ROIs	Line
11/09/22	01_01	1.778	85	ai35hVOS/Scnn1a-tg3-Cre
	02_01	1.466	85	ai35hVOS/Scnn1a-tg3-Cre
	03_01	1.737	85	ai35hVOS/Scnn1a-tg3-Cre
	04_01	1.592	85	ai35hVOS/Scnn1a-tg3-Cre

Slice 1 Loc 1 ROI Occurrence heatmap

download

Max occurrences: 7.0	Pixel	Occurrences
37 64	5.0	25.0%
40 67	5.0	25.0%
38 66	6.0	30.0%
39 69	6.0	30.0%
38 65	6.0	30.0%
42 66	6.0	30.0%
35 62	6.0	30.0%
36 67	7.0	35.0%
37 67	7.0	35.0%
37 68	7.0	35.0%

Slice 3 Loc 1 ROI Occurrence heatmap

download

Max occurrences: 4.0	Pixel	Occurrences
26 12	4.0	20.0%
30 66	4.0	20.0%
32 60	4.0	20.0%
34 49	4.0	20.0%
53 2	4.0	20.0%
22 37	4.0	20.0%
16 66	4.0	20.0%
36 39	4.0	20.0%
37 66	4.0	20.0%
67 38	4.0	20.0%

john-judge commented 1 year ago

How often does each pixel occur in an ROI?

Frequently selected pixels are more likely to be part of a cell. In addition to giving us a selection to focus on for LTP time courses, finding these pixels gives us an idea of how likely cells are to respond (response variability) which can be compared to literature values.

11-04-22

Slice 3 Loc 1 ROI Occurrence heatmap

Max occurrences: 5.0	Pixel	Occurrences
40 48	3.0	15.78%
66 22	3.0	15.78%
29 41	4.0	21.05%
12 5	4.0	21.05%
9 44	4.0	21.05%
20 29	4.0	21.05%
21 52	4.0	21.05%
2 68	4.0	21.05%
46 22	4.0	21.05%
24 71	5.0	26.31%

john-judge commented 1 year ago

Persistent ROIs for df/f Time Courses in LTP experiments

11/04/22 Data (kv2.1hVOS3.0/PV-Cre)

Still no evidence of any LTP. In fact, maybe evidence of LTD. If this is the case, we expect to see the opposite with Scnn1a-tg3-Cre LTP experiments -- 11/09/22 was one such prime example.

11/04/22 Data (ai35hVO/Scnn1a-tg3-Cre)

download

Compare LTP time courses to ROI persistence maps. Is there a correlation between spatial pattern spread and apparent LTD?

john-judge commented 1 year ago

Response Variability: Compare observed to literature

Above, we are observing maximum response percentages of 20% to 35% ROI-selections of stimulations. This may depend on stimulation amplitude relative to spike threshold, as well as how well we are recording and processing the data.

Due to failure and sparsity of synapses, response to stimuli may evoke different patterns on repetition. Huang et al 2022 use Poisson distributions to model response variability of barrel cortices from thalamic input.

They found 0.02-2.2 spikes/stimulus/cell in the network, broken down by neuronal classes:	Neuron class	membrane depolarization state (mV)
Excitatory L4	-80	0.06 +/- 0.11
Inhibitory L4	-80	0.68 +/- 0.71
Excitatory L4	-60	0.44 +/- 0.30
Inhibitory L4	-60	2.13 +/- 1.48

Each neuron has a principal whisker, but typically has multiple whiskers in its receptive field due to thalamic input and cross-columnar projections. Surround-whisker responses are delayed, weaker, and vary more widely with membrane state, i.e. membrane depolarization raises excitability in L2/L3 to modulate cross-columnar integration. Behaviorally, this may affect attention or active sensing / perception of the animal (Huang et al 2022).

john-judge commented 1 year ago

Summary

summary across all occurrence LTP time courses made so far. Only for slice/loc with 15+ recordings.	Cell type	Line	Mean max occurrence %	Std. dev	N (slices)
Pyramidal L4	Scnn1a-tg3-Cre	42%	14%	6
PVI	PV-Cre	34%	10%	6