Error in tf_to_gene about scenicplus HOT 6 CLOSED

Hi @MatthewTCManion

Could you try running the following command in the terminal?

scenicplus grn_inference TF_to_gene \
            --multiome_mudata_fname ACC_GEX.h5mu \
            --tf_names tf_names.txt \
            --temp_dir /data/PetrosLab/Matt/scenicplus/consensus_peak_bulk_750bp/tmp/ \
            --out_tf_to_gene_adjacencies tf_to_gene_adj.tsv \
            --method GBM  \
           --n_cpu 20  \
           --seed 666

Best,

Seppe

from scenicplus.

Hi Seppe, having this exact same error now! Upon running your suggested line I get the exact same error message as running the whole SCENIC+ (screenshot attached)

(EDIT: just realised this is slightly before in the pipeline (region_to_gene instead of tf_to_gene)
**(EDIT #2 (solution): downgrading to python 3.11.8 (was 3.11.9 prior) solved all these issues...) @SeppeDeWinter maybe worthwhile specifying the 3.11.8 and not just 3.11 in the tutorials? :) **

from scenicplus.

Hi Seppe, having this exact same error now! Upon running your suggested line I get the exact same error message as running the whole SCENIC+ (screenshot attached) (EDIT: just realised this is slightly before in the pipeline (region_to_gene instead of tf_to_gene) **(EDIT #2 (solution): downgrading to python 3.11.8 (was 3.11.9 prior) solved all these issues...) @SeppeDeWinter maybe worthwhile specifying the 3.11.8 and not just 3.11 in the tutorials? :) **

I am running it now with the command above, but I had previously tried it with downgrading python 3.11.8 and gotten the same result

from scenicplus.

@SeppeDeWinter Running the command without Snakemake, I get a segfault, but I can't tell why. Usually that's a resource allocation issue, but the memory use stays firmly under the allocated limit.

UPDATE: I tried it again with a larger CPU allocation and it finished correctly, I will test the rest of the pipeline now

from scenicplus.

I had a similar exit on eGRN_extended, but it ran fine when I used the command outside of the Snakemake pipeline:

> scenicplus grn_inference eGRN \
	--is_extended \
	--TF_to_gene_adj_fname tf_to_gene_adj.tsv \
	--region_to_gene_adj_fname region_to_gene_adj.tsv \
	--cistromes_fname cistromes_extended.h5ad \
	--ranking_db_fname /data/PetrosLab/Matt/scenicplus/Nkx_750bp.regions_vs_motifs.rankings.feather \
	--eRegulon_out_fname eRegulons_extended.tsv \
	--temp_dir /data/PetrosLab/Matt/scenicplus/consensus_peak_bulk_750bp/tmp/ \
	--order_regions_to_genes_by importance \
	--order_TFs_to_genes_by importance \
	--gsea_n_perm 1000 \
	--quantiles 0.85 0.90 0.95 \
	--top_n_regionTogenes_per_gene 5 10 15 \
	--top_n_regionTogenes_per_region \
	--min_regions_per_gene 0 \
	--rho_threshold 0.05 \
	--min_target_genes 10 \

I'm not sure what the issue is, but it appears to be with the snakemake workflow, not the specific steps

UPDATE: the snakemake workflow worked for AUCell_extended after running the previous 2 steps manually, but failed on eGRN_direct:

scenicplus grn_inference eGRN
--TF_to_gene_adj_fname tf_to_gene_adj.tsv
--region_to_gene_adj_fname region_to_gene_adj.tsv
--cistromes_fname cistromes_direct.h5ad
--ranking_db_fname /data/PetrosLab/Matt/scenicplus/Nkx_750bp.regions_vs_motifs.rankings.feather
--eRegulon_out_fname eRegulon_direct.tsv
--temp_dir /data/PetrosLab/Matt/scenicplus/consensus_peak_bulk_750bp/tmp/
--order_regions_to_genes_by importance
--order_TFs_to_genes_by importance
--gsea_n_perm 1000
--quantiles 0.85 0.90 0.95
--top_n_regionTogenes_per_gene 5 10 15
--top_n_regionTogenes_per_region
--min_regions_per_gene 0
--rho_threshold 0.05
--min_target_genes 10
--n_cpu 20

from scenicplus.

All steps ran correctly when I set the number of CPUs to 50, it seems it was all just a resource allocation issue.

from scenicplus.