cant make static tSNE plot for data set "ARKctxDevo400sc"

[carlocolantuoni]

i am trying to create a static tSNE scatter for the data sets "ARKctxDevo400sc" but the option does not come up in the pull down menu - what does the system check to enable this? I am currently assembling views in the CarloTEMP profile under curator.

[jorvis]

You said 'data sets' but I only see one with that in the title:

accafd8b-0354-4602-8626-4b5f5ada1d4f - ARKctxDevo400sc

It looks for an analysis where there's a stored tSNE. I don't see that in this dataset. Here's a list of the stored attributes:

---

---

Observation columns: CellID Cell Anatomical.Source Age LibrarySize..Read.Pairs..Million. MappedPairs AlignmentRate..Pairs GenesTagged GW X region donor colorBYdonor colorBYregion pchBYdonor pchBYregion IDnum colorBYdcx COLORby.SLA COLORby.SLC1A3 COLORby.EOMES COLORby.GAD1 COLORby.DCX CellClass PrevSeq COLORbyPrevSeq COLORbyCellClass COLORby.CRYAB COLORby.PDGFD COLORby.TAGLN2 COLORby.PALLD COLORby.HOPX COLORby.PTPRZ1 COLORby.PTN COLORby.TNC COLORby.FAM107A COLORby.MOXD1 XbyCellClass XbyCellClassJ pchBYdonorNregion PC1%var6.14 PC2%var1.79

[carlocolantuoni]

Yes, there is only the 1 ARKctxDevo400sc data set.

What we are currently referring to as a "tSNE" plot is actually just a scatter of any 2 values. in this case i want to use the PCs in the sample meta data to scatter the samples. UMAP will also be used a lot in this way. It might be easier to drop the auto checks and just let the user ask for that visualization when they need it - they could use tSNE, PCA, ICA, UMAP or other methods, so an ahto check is hard.

On Fri, Mar 13, 2020, 16:03 Joshua Orvis notifications@github.com wrote:

You said 'data sets' but I only see one with that in the title:

accafd8b-0354-4602-8626-4b5f5ada1d4f - ARKctxDevo400sc

It looks for an analysis where there's a stored tSNE. I don't see that in this dataset. Here's a list of the stored attributes:

---

Observation columns: CellID Cell Anatomical.Source Age LibrarySize..Read.Pairs..Million. MappedPairs AlignmentRate..Pairs GenesTagged GW X region donor colorBYdonor colorBYregion pchBYdonor pchBYregion IDnum colorBYdcx COLORby.SLA COLORby.SLC1A3 COLORby.EOMES COLORby.GAD1 COLORby.DCX CellClass PrevSeq COLORbyPrevSeq COLORbyCellClass COLORby.CRYAB COLORby.PDGFD COLORby.TAGLN2 COLORby.PALLD COLORby.HOPX COLORby.PTPRZ1 COLORby.PTN COLORby.TNC COLORby.FAM107A COLORby.MOXD1 XbyCellClass XbyCellClassJ pchBYdonorNregion PC1%var6.14 PC2%var1.79

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/nemoarchive/analytics/issues/100#issuecomment-598890207, or unsubscribe https://github.com/notifications/unsubscribe-auth/AH7KC7T3P7U5MIVN4A4X6A3RHKGQFANCNFSM4LFX3K2A .

[carlocolantuoni]

@jorvis - can you make a static scatter plot an option for all data sets? perhaps in addition to the 2D static scatter you already have, can we also make static scatters such as age or time (selected by user) on the x-axis, and expression on the Y? i ask as the dynamic plots generated now use only about 30% of the space in there box and are often hard to see.

[carlocolantuoni]

How can i make a static scatter plot for this data set?

[jorvis]

See if we can demo and do static tSNE with these columns.

PC1%var6.14 PC2%var1.79

Investigate removing the naming convention entirely.

[jorvis]

Is this the sort of view you wanted? Screenshot using the current curator to generate a scatter using those columns:

[carlocolantuoni]

thats the closest i could get to the static tSNE - type plot, so i used that 1, but the tool u have to make static tSNEs with the colored "legend" panel next to it is really what id like here

On Thu, Jun 18, 2020 at 1:56 PM Joshua Orvis notifications@github.com wrote:

Is this the sort of view you wanted? Screenshot using the current curator to generate a scatter using those columns:

[image: Screenshot from 2020-06-18 12-55-37] https://user-images.githubusercontent.com/330899/85055418-204cbc80-b163-11ea-876c-3a258936c692.png

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/nemoarchive/analytics/issues/100#issuecomment-646218052, or unsubscribe https://github.com/notifications/unsubscribe-auth/AH7KC7WSLKT73Q74TEDJNS3RXJINZANCNFSM4LFX3K2A .

-- Carlo

[carlocolantuoni]

Hey joshua, I have a number of datasets like this that i need to use the static tSNE for - let me know wen the column name checks are off so i can build those - thanks Carlo

On Thu, Jun 18, 2020, 15:12 Carlo Colantuoni colantuonicarlo@gmail.com wrote:

thats the closest i could get to the static tSNE - type plot, so i used that 1, but the tool u have to make static tSNEs with the colored "legend" panel next to it is really what id like here

On Thu, Jun 18, 2020 at 1:56 PM Joshua Orvis notifications@github.com wrote:

Is this the sort of view you wanted? Screenshot using the current curator to generate a scatter using those columns:

[image: Screenshot from 2020-06-18 12-55-37] https://user-images.githubusercontent.com/330899/85055418-204cbc80-b163-11ea-876c-3a258936c692.png

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/nemoarchive/analytics/issues/100#issuecomment-646218052, or unsubscribe https://github.com/notifications/unsubscribe-auth/AH7KC7WSLKT73Q74TEDJNS3RXJINZANCNFSM4LFX3K2A .

-- Carlo

[jorvis]

I looked into it more to remind myself and there actually aren't column checks at that point at all. The curator is looking for a stored tSNE analysis (stored in h5ad as an obsm['X_tsne'] entry) and if present it displays that as an option.

What I do have though is a utility script created which will accept any column name pairs I give it and create such an analysis, so I could use that to make sure all your datasets get that analysis entry created. So what I need from you is either the list of column names you're going to create (like the "PC1%var6.14" and "PC2%var1.79" pair above) or you could create a consistent name on the future datasets. Either is fine.

In the future we can look into letting users do this at the curation level but that's engineering I don't have time to do before next week.

[carlocolantuoni]

I have created several static tSNEs from datasets without a stored tSNE and it let me pick whatever columns i wanted. See CarloTEMP profile under NeMO curator.

Can u chat now?

On Fri, Jun 19, 2020, 23:38 Joshua Orvis notifications@github.com wrote:

I looked into it more to remind myself and there actually aren't column checks that point at all. The curator is looking for a stored tSNE analysis (stored in h5ad as an obsm['X_tsne'] entry) and if present it displays that as an option.

What I do have though is a utility script created which will accept any column name pairs I give it and create such an analysis, so I could use that to make sure all your datasets get that analysis entry created. So what I need from you is either the list of column names you're going to create (like the "PC1%var6.14" and "PC2%var1.79" pair above) or you could create a consistent name on the future datasets. Either is fine.

In the future we can look into letting users do this at the curation level but that's engineering I don't have time to do before next week.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/nemoarchive/analytics/issues/100#issuecomment-646932333, or unsubscribe https://github.com/notifications/unsubscribe-auth/AH7KC7QTYW2YMGZ6B7K56NTRXQVKZANCNFSM4LFX3K2A .

[jorvis]

Commit 39505290 addresses this. Re-open if there are issues.