doublet detection errors with troublet: non-zero exit status and index out of bounds #231
Comments
|
Probably upstream of doublet detection if len is 0. Can you give me the first few lines of alt.mtx and the cluster_tmp.tsv? |
|
First lines of alt.mtx But cluster_tmp.tsv is empty |
|
Right so u are using the unfiltered barcodes file. Only use barcodes that are cells according to cellranger. This file should be in outs/filtered/ or something like that |
|
The filepath is going to the filtered ones for each sample but maybe something got corrupted in them. I'll try some different samples to see if they will run and I will try taking a closer look at the filtered barcodes files that aren't working. |
|
Well that alt.mtx says you have 6.79 million cells |
|
That doesn't sound right at all but I'm looking into it. Thanks for your help in figuring out where I need to look for troubleshooting these runs! |
|
I'm having this same exact error. Did you ever figure out how to fix it? |
|
Use the filtered barcodes not the raw barcodes |
|
In cellranger output there is a filtered folder and raw folder. Get it from the filtered folder |
|
Hello thank you for your quick reply. My barcodes.tsv file is from the filtered folder, and I still got those errors. Do you know any other possibilities that could be causing this error? |
|
Give me the first 3 lines of alt.mtx |
|
My first few lines of alt.mtx %%MatrixMarket matrix coordinate real general |
|
Right. This is due to an unflitered barcodes file. This says u have 2 million cells |
|
I am also having the same error. I am using the filtered barcode. $ head ref.mtx |
|
Okay, this looks like an actual bug. It seems to be caused when at least one cell has 0 genetic variants. Normally this is not the case, but its possible. Can I get a wc -l on the barcodes.tsv? Will help me pin down the issue to know if its 20612 as above or a diff number. |
|
zcat -c ../outs/filtered_feature_bc_matrix/barcodes.tsv.gz |wc -l |
|
and is there any output in clusters.err? I assume that clusters_tmp.tsv is empty -- that is the cause of this error. But I'm not sure why it would be empty unless souporcell fails silently. |
|
No they are not empty. head clusters.err |
|
oh, interesting. and the first lines of clusters_tmp.tsv? maybe all unassigned? |
|
I have 84 genotyped individuals. I think it is assigned. |
|
seems that many clusters are exactly the same. How that causes this error I'm still figuring out. I guess if some clusters are the same as other clusters, no cells will be assigned to them. But even if I fix that, we will have another more difficult problem. One is a problem I'm actively researching -- clustering many individuals. Previously we have tested up to 21 individuals and that only works with many random restarts and a high number of umi/cell. 84 is definitely out of the scope of the current software. Clustering is an NP hard problem and this is additionally difficult because it is sparse data and binomial not euclidean. We should move this to an email thread if you want to continue this conversation. whheaton@gmail.com |
|
I see. Thanks for very quick responses and clarification. Really appreciated. |
I completed the tutorial and two samples of my own without error but now I'm consistently getting this problem with subsequent samples:
I checked the doublets.err file which says:
But I'm not sure how to correct this issue. Any advice or things to try is appreciated. Thank you!
The text was updated successfully, but these errors were encountered: