search

ParkerICI / vite

R package for analysis of single-cell data using graphs
GNU General Public License v3.0
10 stars 4 forks source link

Error in `[.data.frame`(tab, , col.names) : undefined columns selected #5

Open luglilab opened 5 years ago

luglilab commented 5 years ago

Hi,

I'm trying to execute "run_scaffold_analysis" function but I got the following error message:

Processing /XXXX/XXX/XX/Tissue.pooled.clustered.txt Running with Edge weight: 7.000000 Error in [.data.frame(tab, , col.names) : undefined columns selected

Could you give some tips to solve this problem. In the text file you will find my code. Input FCS file coming from facs symphony.

vite.R.txt

rkageyama commented 5 years ago

Hi lugliab, Not sure off the bat, but it could be something to do with the column names. They are listed in the format Comp.APC.A....CXCR3" which might be off from what are in the fcs files?

Could you share some of the fcs files that are causing this error? (via box or dropbox)

luglilab commented 5 years ago

Hi rkageyama,

thanks for reply. Here you can find the fcs https://www.dropbox.com/sh/uszww7xfc1obomr/AABLTb2qa8ZlDvyxe4aYQXCUa?dl=0

FCS 3.0 files were imported into FlowJo software version 9, and analysed by standard gating to remove aggregates and dead cells, and identify CD4+ T cells. T cells were subsequently imported in FlowJo version 10 and biexponentially transformed.

Some markers are already excluded in the file name you will find blood tumor and tissue that are groups that i have used for clustering.

If you need others information, let me know.

Best regards,

Simone

rkageyama commented 5 years ago

Hi Simone, Thanks for sharing the fcs. Would you mind also sharing the clustering files? That should hopefully let us reproduce the error. Thanks! -Robin

luglilab commented 5 years ago

Hi,

I hope this cluster file is fine.

Simone

clusters_k75.xlsx

rkageyama commented 5 years ago

Hi Simone, I tried running some of your fcs files with the col.names you had written in the script you sent, and it seemed to work. It looks like vite is having a hard time matching the col.names with the channels in the fcs files, so I was wondering if the cytofCore or cytof_exprsMerge could be messing with the channel names of the fcs files. Do the fcs files in the landmarks folder after transformation/concatenation/subsetting still have the same channel names as the originals?

luglilab commented 5 years ago

Dear Robin,

you've right, there's a mistake in the col.name Ki-67 (wrongly written in some fcs). We have fixed this problem and update the files in the dropbox folder. Further, I have attached in the same directory the file with the markers that we excluded for the clustering (Markers_to_exclude.txt) while the cluster file seems correct.

Sorry for the inconvenience

Best regards,

Simone