QC1 and QC4: Run on values per sweep?

[MichaelClerx]

QC1 and 4 inspect Rseal, Cm, Rseries "before" and "after".
Other QCs look for stability between "sweeps"

If "sweeps" are actual sweeps (machine repeats), then we might have different values per sweep and so could/should use these?
If they're always the same, e.g. because the machine doesn't re-estimate or the "sweeps" are really just different staircases in the same long protocol, then it makes sense as it is...

@chonlei @joeyshuttleworth @mirams can you shed some light on this?

[MichaelClerx]

(Related: maybe we should avoid the word sweep)

[mirams]

Well the original paper there were two 'sweeps' of the staircase before and two sweeps after drug (marked with 'C'):

(more recent Joey stuff has two staircases 'bookending' all the other protocols - eight in total)

We phrase QC1 in the original paper as "through the whole set of measurements" - I don't see any reason why we shouldn't be applying it to every machine sweep of every protocol (before and after full block/drug addition). It's generally set to a 'bare minimum' sort of value on the actual machine software, but we can override it in our postprocessing. QC1 could change a lot before / after the 5 mins of drug addition (especially when squirting in the drug can dislodge the cell) and we think a lot of cells might have dropped off or suchlike.

I would imagine QC3 should check the current between these consecutive sweeps looks pretty much the same, and we are in a 'stable' situation. i.e. QC3 should be used to compare when we are expecting to record the same thing.

[mirams]

On the terminology, I'm not actually sure whether machine re-estimates Rseal etc. on sweeps (e.g. Protocol 1 above has lots of test potentials so Protocol 1 has 7 sweeps in it in conventional e-phys terminology) or on Protocols as drawn above. We should figure that out. It is probably on Protocols in which case we should avoid 'sweep', and use 'protocol repeat' (which I was not above)!

[joeyshuttleworth]

I've made it so QC1 and QC4 are run on each sweep. This has changed the test cases a little bit because 6/7(?) extra wells are failing where they weren't before in test_run_qc.

On the branch for this I've changed the test cases. But these changes might affect other analysis
https://github.com/CardiacModelling/pcpostprocess/blob/6c4302246d96ea0abb0442a880b2361f9f53d16d/tests/test_herg_qc.py#L893C1-L927C1

[MichaelClerx]

Oh!
Are we updating things already then?
In my mind, we'd made a Lei2019QC or something, and then later add variations 😅

[joeyshuttleworth]

I'm not sure if we want to update this or not. But I want to check if it makes an important difference to my QC output. Looks like it might in some cases.