I tried defining scplus_genes before the line of code you highlighted. So far it appears to have solved the issue.

from scenicplus.

Thank you! That seemed to work but now I am getting a new error

from scenicplus.

Thank you! That seemed to work but now I am getting a new error

Okay, concatenate error solved for my part. I am just going to assume you processed your RNA-seq data in Seurat.
I did this to fix the issue. It was damn near impossible to find the source of the error, but in the end it turned out to be an issue with how seurat exports h5ad files. For some reason it doesn't save the gene names in the right place, which means that the ACC_GEX.h5mu file you generate won't have the gene names stored correctly either.

import scanpy as sc

adata = sc.read("rna.h5ad")
new_adata = adata.raw.to_adata()
new_adata.var_names = adata.raw.var['_index'] #Or whatever your index is called
new_adata.var_names.name = None
new_adata.raw = new_adata
sc.pp.normalize_total(new_adata, target_sum=1e4)
sc.pp.log1p(new_adata)
new_adata.write("rna_new.h5ad")

And then just use the new file in your config script.

Hope this works for you too. If so, then your next error will probably be this:

If you do get this error, updating dask to 2024.4.1 seems to fix it even though it warns you that it is not compatible with scenicplus

Also side note. If my assumption that you worked with seurat is correct, you may also have to go back into R and make sure to delete scale.data, because:

Add assay data

assay.group <- source[['assays']][[assay]]
if (source$index()[[assay]]$slots[['scale.data']]) {
x.data <- 'scale.data'
raw.data <- 'data'
} else {
x.data <- 'data'
raw.data <- if (source$index()[[assay]]$slots[['counts']]) {
'counts'
} else {
NULL
}
}

The data being called "raw.data" is not actually raw
(Snippet from seurat-disk convert.R

from scenicplus.

Hmmm that sounds strange

I literally just did this:
pip install dask==2024.4.1

And ran it again and then it worked

from scenicplus.

Hi @orian116

This is an issue with your genome annotation file (this should contain a column named "Gene").

Can you show the head of genome_annotation.tsv

All the best,

Seppe

from scenicplus.

Hi @SeppeDeWinter
Here is the head and tail of my genome annotation file. It does have the 'Gene' column:

from scenicplus.

I have this exact same issue and also do have the "Gene" column in my gene_annotation.tsv

from scenicplus.

@SeppeDeWinter I have narrowed down the error a bit. It seems like in the gene_search_space.py function, the column names are present before running pr_gene_annotation = pr.PyRanges(gene_annotation.query("Gene in @scplus_genes").copy()) but not after. I will try converting to a pyranges object without this. What file is "@scplus_genes" referencing? Here's a snippet of the code.

from scenicplus.

Yea same here. No objects to concatenate... Stuck on the same thing.

from scenicplus.

Thank you! That seemed to work but now I am getting a new error

Okay, concatenate error solved for my part. I am just going to assume you processed your RNA-seq data in Seurat. I did this to fix the issue. It was damn near impossible to find the source of the error, but in the end it turned out to be an issue with how seurat exports h5ad files. For some reason it doesn't save the gene names in the right place, which means that the ACC_GEX.h5mu file you generate won't have the gene names stored correctly either.

import scanpy as sc

adata = sc.read("rna.h5ad") new_adata = adata.raw.to_adata() new_adata.var_names = adata.raw.var['_index'] #Or whatever your index is called new_adata.var_names.name = None new_adata.raw = new_adata sc.pp.normalize_total(new_adata, target_sum=1e4) sc.pp.log1p(new_adata) new_adata.write("rna_new.h5ad")

And then just use the new file in your config script.

Hope this works for you too. If so, then your next error will probably be this:

If you do get this error, updating dask to 2024.4.1 seems to fix it even though it warns you that it is not compatible with scenicplus

Also side note. If my assumption that you worked with seurat is correct, you may also have to go back into R and make sure to delete scale.data, because:

Add assay data

assay.group <- source[['assays']][[assay]] if (source$index()[[assay]]$slots[['scale.data']]) { x.data <- 'scale.data' raw.data <- 'data' } else { x.data <- 'data' raw.data <- if (source$index()[[assay]]$slots[['counts']]) { 'counts' } else { NULL } }

The data being called "raw.data" is not actually raw (Snippet from seurat-disk convert.R

How did you get to upgrade dask without upgrading pandas? When I run scenicplus after following your steps, I get an error for is_categorical being deprecated

from scenicplus.