-
Notifications
You must be signed in to change notification settings - Fork 9
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Enhance adaptive suprathreshold #2302
base: main
Are you sure you want to change the base?
Conversation
daec7e1
to
886e9e6
Compare
d7b9af8
to
7a4e927
Compare
87bc99b
to
03ebd15
Compare
This comment was marked as outdated.
This comment was marked as outdated.
c643b38
to
eaa0825
Compare
00b6e6f
to
ae1e170
Compare
@timjarsky Ready to test. |
ae1e170
to
17d9733
Compare
@timjarsky Investigation of the live testing from 2025-01-28. H25.26.401.11.53.01.01.pxpNo adaptive data Sst-IRES-Cre.pxp34-49: Passing f-I slope QC and sweep QC on sweeps 46, 48 and 49. 58-75: Passing f-I slope QC and sweep QC on sweeps 72-74. Tricky to see due to +1 offset.
Sst-IRES-Cre (1).pxp10-26: Passing f-I slope QC and sweep QC on sweeps 24, 25 and 26. 37-52: Passing f-I slope QC and sweep QC on sweeps 49-51. 59-68: Passed because the inital FI slope was invalid because 2x the same DAScale. ??? 83-99: Passed because 97: f-I slope QC 99 is the pos/neg slope fillin, but it has the wrong DAScale value. ???
Sst-IRES-Cre (2).pxp18-24: pos/neg slope on 23 and 24. Missing fillin due to bug. ??? 35-44: passing f-I slope on 42-43 (both from RhSuAd) and 44
Sst-IRES-Cre (3).pxp44-59: Sweeps 57 and 58 have f-I neg slope QC. 59 is fillin with wrong dascale ??? 74-84: Passed because the inital FI slope was invalid because 2x the same DAScale. ???
|
We need to check each given labnotebook key where the potential wildcard matches. Bug introduced in d301352 (SweepFormula labnotebook: Enhance it, 2024-09-10).
This avoids having to copy, clean and duplicate all the labnotebook keys for each search key.
… feature/2302-another-adaptive-revision * bugfix/2346-labnotebook-sweepformula-bug-wildcard: SFH_OperationLabnotebookExpandKeys: Prefer FindIndizes SF labnotebook: Fix multiple keys with wildcards
For a future application we want to check for something different than equality, so making the predicate configurable here is required.
…PSQ_FMT_LBN_SWEEP_PASS This has a different explanation but in this sense it is never used. Introduced in c53b5d0 (PSQ_DaScale: Enhance adaptive mode, 2024-07-26).
Having a duplicate version for PSQ_DAScale Adaptive makes it easier to check if we have all variants (offsets, RhSuAd, DAScale, etc.) properly done.
…out baseline QC Needed for a future application.
The stock Igor Pro functions don't maintain the relative order of the same key values. So we had to roll our own.
…terFrequencyCurrentData
In that way this is not dead code in the tests. It is also clearer that we check this error condition first.
- Ignore sweeps with negative fI slopes for the DAScale estimation - Store a new QC value if the fI slope is finite and negative - Using this new QC we are now also done if: - We have to consecutive negative fI slope values (as usual failing sweep QC sweeps can be in between) - or we have a passing fI slope and a consecutive (as above) negative fI slope value If we finish due to these two conditions we also fillin another DAScale in between the positive FI slope sweep and the first negative fI slope value. - We now use the new analysis parameter DAScaleNegativeSlopePercent, defaulting to 50%, to calculate a new DAScale value after a negative slope. We use the percentage of the maximum DAScale step for that. - Add a new labnotebook entry which denotes if a value is filled in or not - Also a bug is fixed which did not ignore fillin sweeps when trying to determine if the set is passing. - And similiar bug could result in fillin sweeps being marked as fI slope passing.
In (PSQ_DAScale: Change fitting logic and drop NumPointsForLineFit, 2024-07-10) we added dropping duplicated AP frequencies in the input data for fitting. We do catch that now with the neg fI slope QC better. And that also removes the hack to find a good DAScale in case this was filtered out.
This is also expected in the code.
We need to search for a change of the f-I neg slope QC, so going from 0 to 1. Unfortunately FindSequence has a bug when reverse searching, so we need to workaround that.
17d9733
to
82df058
Compare
H25.03.302.11.53.01.03.pxp:
|
Close #2258
Next round after 7739d31 (PSQ_DS_FitFrequencyCurrentData: Remove it, 2025-01-28):