Find marker genes step returns errors on a particular dataset

[beamilon]

Sorry if I missed something and it is being taking care of, but the 'find marker genes' from the primary analysis for P25-P27, mouse, scRNA-seq, cochlear sensory neurons (Goodrich) returns an "Error reporting marker genes"

[songeric1107]

this needs to be re-uploaded, there is no umap coordinate in the observation file, the primary function will not work.

[songeric1107]

201

[songeric1107]

reupload the dataset again with tsne coordinate.

https://umgear.org/dataset_curator.html#/dataset/53f99080-29cd-44ca-4c14-b26ca1ad7588/displays

[beamilon]

The primary is still not working. Since 221 was closed, I thought the dataset had been re-uploaded. Is there a different problem?

[jorvis]

Is this still an issue? I can't see this dataset when I check the single-cell tool, though there is another whose name is very close (doesn't have the ", TSNE" part within the name).

[beamilon]

The dataset in the public profile is the one without "TSNE": P25-P27, mouse, scRNA-seq, cochlear sensory neurons (Goodrich) There is a primary analysis available apparently but the 'Find marker genes' returns an error message.

[songeric1107]

@beamilon , the TSNE coordinate only adds to the one with TSNE in the title, the other one is for violin plot /svg purposes.

If you want to use the one with tsne, we should replace the one in the profile with the new one. without TSNE: c32835d3-cac4-bb0e-a90a-0b41dec6617a new with TSNE:53f99080-29cd-44ca-4c14-b26ca1ad7588

[beamilon]

@songeric1107 Yes, I think we should replace with the new one to have access to the primary analysis

[songeric1107]

@beamilon @jorvis , this dataset is included in multiple profile, what is the best way to replace it.

scRNA-seq / Cell Types/Tissues / Andy / Neuron / SGNs / Neuron Total / Neuron / PMB SGNs / scRNAseq+deafenedRats / Tom / My Profile FW / test / single cell sq (interest) / Sian test / MyProfile01 / Neuron / Mouse cochlea Single-c RNAseq / SGN RNAseq c32835d3-cac4-bb0e-a90a-0b41dec6617a P25-P27, mouse, scRNA-seq, cochlear sensory neurons (Goodrich) single-cell-rnaseq rhertzano@smail.umaryland.ed30078709 GSE114997 Public 2019-01-31 18:37:40

[jorvis]

So is that the only thing which has changed? tSNE was added but everything else stayed the same?

[songeric1107]

the column names are different, but the plot is the same

[songeric1107]

this dataset is included in multiple profiles, I am not sure whether other users have any curations using the old version.

@beamilon any comments?

[beamilon]

@songeric1107 That's a good point. I will ask Ronna.

[jorvis]

I'll bring convo back here. I checked and 5 different users have created curations for this dataset. If I can have a full list of the column names which have changed I can see which will break and correct them.

[songeric1107]

• old version without TSNE (in the public profiles):c32835d3-cac4-bb0e-a90a-0b41dec6617a,

3 optional columns names for plotting: position(Apex,Base,Middle), tissue_type(1_A,1_B,1_C,T2), cluster_labe( Apex_1_A, Apex_1_B........Middle_1_A,Middle_1_B,Middle_1_C)

• new version with TSNE ():53f99080-29cd-44ca-4c14-b26ca1ad7588. 3 columns except for the tsne coordinate: condition(Apex, Base,Middle)---maybe change this as "position" cell_type(A_T1,B_T1,C_T1,T2) cluster_label(A_T1,B_T1,C_T1,T2)

[jorvis]

OK, so going from old to new, the changes are:

• position -> condition
• tissue_type -> cell_type

And I'm assuming cluster_label hasn't changed (or was it actually misspelled the first time as 'cluster_labe' ?

[songeric1107]

the cluster_label is different, the old one is the combination of the position+tissue_type; the cluster_label of the recent one is the same as the cell_type.

[jorvis]

So the values within that column are different but the column name is the same. I think think there's anything to do with that one then.

[jorvis]

I've replaced the h5 files with the new version and updated all the layout configs to reflect the new column names. Verified that displays still work:

And that marker genes now work:

[jorvis]

The 53f99080-29cd-44ca-4c14-b26ca1ad7588 dataset can be deleted.