Can't use the cell type labels for the mpfc experiment

[jules-samaran]

Hi, I am trying to use the data available at https://www.dropbox.com/sh/f7ebheru1lbz91s/AACK9iPT11K13PYck7uCUrvXa/mpfc_160?dl=0&subfolder_nav_tracking=1 to harmonize different datasets including yours. However when I tried using the cell_types for the 20180417_BZ5_control experience, I noticed that inside the class_labels.csv file for each cell we are given a cluster ID and a cluster name but they don't make sense. For example a cell has the cluster id 11 and its cluster name is andro and another cell has the cluster id 11 but has the cluster name endo. There doesn't seem to be any coherence between the cluster ID and the cluster name, some cluster ids are paired with each one of the cluster names at least once when you look at all the cells. Which one is supposed to be the accurate classification? Best, Jules

[weallen]

The cluster names should be correct -- it's what I used to check that the export was working (e.g. https://github.com/weallen/STARmap/blob/master/python/examples/example_load_experiment.ipynb). It looks like the clusterIDs may have been permuted somehow. I'll try to fix the bug and re-export.

[jules-samaran]

Hi, I'm currently trying to use the Visual1020 datasets https://www.dropbox.com/sh/f7ebheru1lbz91s/AADm6D54GSEFXB1feRy6OSASa/visual_1020/20180505_BY3_1kgenes?dl=0&subfolder_nav_tracking=1 but the number of labels in class_labels.csv is 1389 whereas the counts matrix has 1549 rows (cells). Are there cell type annotations I can use as labels for the dataset's cells?

[kimmo1019]

Hi, I'm currently trying to use the Visual1020 datasets https://www.dropbox.com/sh/f7ebheru1lbz91s/AADm6D54GSEFXB1feRy6OSASa/visual_1020/20180505_BY3_1kgenes?dl=0&subfolder_nav_tracking=1 but the number of labels in class_labels.csv is 1389 whereas the counts matrix has 1549 rows (cells). Are there cell type annotations I can use as labels for the dataset's cells?

Hey, have you solved this problem? I'm facing the same...

[soerenab]

Same problem here - how do the 1389 annotations map to the 1549 cells?