`MontSection` and `FrameSet` files are incompatible with single df model

[alisterburt]

Support was recently added for montage section and frameset mdoc files in #10 and #11

In an effort to continue with the 'single dataframe' model I added extra fields to the section data but this means we have two incompatible data types in the dataframe and a lot of nans

It seems like ZValue is used across all types of mdoc files when there are sets of images.
For FrameSet/MontSection files would it be better to return a dict[str, pd.DataFrame] with one df stored under ZValue and one stored at FrameSet/MontSection?

Given that there only ever seems to be one entry for FrameSet/MontSection maybe that table is better as a dict than a dataframe?

Thoughts @jojoelfe @m-albert?

[jojoelfe]

At least for the FrameSet with multiple ZValue, my preference would be to return one DataFrame with rows only for the ZValue sections, but to include the columns from the FrameSet. So this:

	ZValue	FrameSet	TiltAngle	StagePosition	StageZ	Magnification	MagIndex	Intensity	ExposureDose	DoseRate	SpotSize	Defocus	TargetDefocus	ImageShift	RotationAngle	ExposureTime	Binning	UsingCDS	CameraIndex	DividedBy2	LowDoseConSet	PriorRecordDose	SubFramePath	NumSubFrames	DateTime	FilterSlitAndLoss	Voltage	titles
0	nan	0.0	32.986	(-142.831, -457.473)	-13.7203	33000.0	26.0	0.11378	291.302	2.15073	9.0	-7.933	-4.5	(0.0, 0.0)	178.414	40.9854	0.5	False	1.0	False	4.0	nan	X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif	243.0	08-Oct-21 07:47:29	(20.0, 0.0)	300.0	[]
1	0.0	nan	33.0		nan	nan	nan	nan	7.66184	nan	nan	nan	nan		nan	nan	nan		nan		nan	0.0		nan			300.0	[]
2	1.0	nan	33.0		nan	nan	nan	nan	7.66184	nan	nan	nan	nan		nan	nan	nan		nan		nan	7.66184		nan			300.0	[]
3	2.0	nan	33.0		nan	nan	nan	nan	7.66184	nan	nan	nan	nan		nan	nan	nan		nan		nan	15.3237		nan			300.0	[]
4	3.0	nan	33.0		nan	nan	nan	nan	7.66184	nan	nan	nan	nan		nan	nan	nan		nan		nan	22.9855		nan			300.0	[]
5	4.0	nan	33.0		nan	nan	nan	nan	7.66184	nan	nan	nan	nan		nan	nan	nan		nan		nan	30.6473		nan			300.0	[]
6	5.0	nan	33.0		nan	nan	nan	nan	7.66184	nan	nan	nan	nan		nan	nan	nan		nan		nan	38.3092		nan			300.0	[]
7	6.0	nan	33.0		nan	nan	nan	nan	7.55522	nan	nan	nan	nan		nan	nan	nan		nan		nan	45.971		nan			300.0	[]
8	7.0	nan	33.0		nan	nan	nan	nan	7.66184	nan	nan	nan	nan		nan	nan	nan		nan		nan	53.5262		nan			300.0	[]
9	8.0	nan	33.0		nan	nan	nan	nan	7.54812	nan	nan	nan	nan		nan	nan	nan		nan		nan	61.1881		nan			300.0	[]
10	9.0	nan	33.0		nan	nan	nan	nan	7.66184	nan	nan	nan	nan		nan	nan	nan		nan		nan	68.7362		nan			300.0	[]
11	10.0	nan	33.0		nan	nan	nan	nan	7.66184	nan	nan	nan	nan		nan	nan	nan		nan		nan	76.398		nan			300.0	[]
12	11.0	nan	33.0		nan	nan	nan	nan	7.66894	nan	nan	nan	nan		nan	nan	nan		nan		nan	84.0599		nan			300.0	[]
13	12.0	nan	33.0		nan	nan	nan	nan	7.66184	nan	nan	nan	nan		nan	nan	nan		nan		nan	91.7288		nan			300.0	[]
14	13.0	nan	33.0		nan	nan	nan	nan	8.10961	nan	nan	nan	nan		nan	nan	nan		nan		nan	99.3906		nan			300.0	[]
15	14.0	nan	33.0		nan	nan	nan	nan	7.66894	nan	nan	nan	nan		nan	nan	nan		nan		nan	107.5		nan			300.0	[]
16	15.0	nan	33.0		nan	nan	nan	nan	7.66184	nan	nan	nan	nan		nan	nan	nan		nan		nan	115.169		nan			300.0	[]
17	16.0	nan	33.0		nan	nan	nan	nan	7.66184	nan	nan	nan	nan		nan	nan	nan		nan		nan	122.831		nan			300.0	[]
18	17.0	nan	33.0		nan	nan	nan	nan	7.66184	nan	nan	nan	nan		nan	nan	nan		nan		nan	130.493		nan			300.0	[]
19	18.0	nan	33.0		nan	nan	nan	nan	7.66184	nan	nan	nan	nan		nan	nan	nan		nan		nan	138.155		nan			300.0	[]
20	19.0	nan	33.0		nan	nan	nan	nan	7.66184	nan	nan	nan	nan		nan	nan	nan		nan		nan	145.817		nan			300.0	[]

---

becomes this:

	ZValue	FrameSet	TiltAngle	StagePosition	StageZ	Magnification	MagIndex	Intensity	ExposureDose	DoseRate	SpotSize	Defocus	TargetDefocus	ImageShift	RotationAngle	ExposureTime	Binning	UsingCDS	CameraIndex	DividedBy2	LowDoseConSet	PriorRecordDose	SubFramePath	NumSubFrames	DateTime	FilterSlitAndLoss	Voltage	titles
1	0.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66184	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	0.0	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
2	1.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66184	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	7.66184	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
3	2.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66184	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	15.3237	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
4	3.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66184	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	22.9855	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
5	4.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66184	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	30.6473	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
6	5.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66184	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	38.3092	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
7	6.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.55522	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	45.971	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
8	7.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66184	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	53.5262	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
9	8.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.54812	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	61.1881	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
10	9.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66184	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	68.7362	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
11	10.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66184	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	76.398	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
12	11.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66894	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	84.0599	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
13	12.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66184	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	91.7288	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
14	13.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	8.10961	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	99.3906	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
15	14.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66894	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	107.5	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
16	15.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66184	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	115.169	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
17	16.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66184	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	122.831	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
18	17.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66184	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	130.493	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
19	18.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66184	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	138.155	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]
20	19.0	[0.0]	33.0	[-142.831, -457.473]	[-13.7203]	[33000.0]	[26.0]	[0.11378]	7.66184	[2.15073]	[9.0]	[-7.933]	[-4.5]	[0.0, 0.0]	[178.414]	[40.9854]	[0.5]	[False]	[1.0]	[False]	[4.0]	145.817	[PosixPath('X:\Johannes_20211007\grid2_lamella3\frames\s_mmm_01013_33.0_Oct08_07.46.46.tif')]	[243.0]	['08-Oct-21 07:47:29']	[20.0, 0.0]	300.0	[]

---

[m-albert]

@alisterburt thanks for the ping :)

@jojoelfe I might be misunderstanding the design you mention, but wouldn't sections other than ZValue in this case not be able have their own values for shared columns?

I think I like the option of returning a dict[str, pd.DataFrame] regardless of whether it contains or not non-ZValue sections. Because

• it's general yet explicit
• reduces the need to fill in nans and
• represents a minimal change wrt to before (i.e. still providing easy and convenient access to a dataframe containing only ZValues).

[alisterburt]

Thanks for the feedback both! It's great to see what everyone would want

@jojoelfe replicating the data from the MontSection or FrameSet entry across all ZValues seems like the nicest API to me but column names do clash with differing data between the two tables and I'm not 100% on the best way to handle that.

Some options

1. replicate values from the FrameSet section across new columns with a FrameSet prefix
2. same as above, but only add the prefix for clashing names

I think I prefer option 1 of these two as it's less magical and more predictable

@m-albert because the MontSection DataFrame would only ever have one row I think I prefer the merging strategy overall, would solution 1 above be an okay solution for you?

Final question would be on how we handle single FrameSet files (example added by @jojoelfe) - I think saying that ZValue is the special case which gets no prefix and FrameSet/MontSection always get a prefix feels natural/predictable

If we've got consensus here I'll go ahead and implement, unless either of you particularly wants to submit a PR? Happy either way, just let me know

[m-albert]

Solution 1 would definitely be okay with me, thanks for asking :)

[jojoelfe]

Number 1 is definitely ok. I wonder if it would make sense to check if the values are the same for a column in FrameSet/MontSection and ZValues, and then to skip adding a column with prefix. This seems to be the case for a lot of columns in the MontSection files.

because the MontSection DataFrame would only ever have one row I think I prefer the merging strategy overall, would solution 1 above be an okay solution for you?

This is not necessarily true. SerialEM allows for multiple montages in one mrc files. I've submitted PR #14 to add a test for such a case. Not sure what would be the best way there. I still think merging and presenting a flat DataFrame would be best, but I would be happy with any option.

[alisterburt]

brill - I'll have a play around this weekend then! Thanks again both