Error in bash

[Khalid-Usman]

I have just configured the conda and when I try to run evaluation on GPU, I get the following error.

Your guidance in resolving this issue will be highly appreciated and if you guide me to run only GLUE or 2,3 methods instead of all. I tried to comment on other methods in the config file but seems like it's not working.

Thanks

[Jeff1995]

Hi @Khalid-Usman! Which lines did you comment out in the config file? There are multiple types of benchmarks in the config file, you may try commenting out all but one benchmark (

GLUE/evaluation/config/config.yaml

Lines 182 to 333 in fc6c209

	benchmark_noprep:
	data:
	prior:
	gene_region: combined
	extend_range: 0
	corrupt_rate: 0.0
	corrupt_seed: 0
	method:
	UnionCom:
	Pamona:
	MMD_MA:
	iNMF:
	iNMF_FiG:
	LIGER:
	LIGER_FiG:
	Harmony:
	bindSC:
	CCA_anchor:
	GLUE:
	dim: 50
	alt_dim: 0
	hidden_depth: 2
	hidden_dim: 256
	dropout: 0.2
	lam_graph: 0.02
	lam_align: 0.05
	neg_samples: 10
	seed: 8
	#------------------------------ Prior robustness -------------------------------
	prior_corrupt:
	data:
	prior:
	gene_region: combined
	extend_range: 0
	corrupt_rate:
	- 0.2
	- 0.4
	- 0.6
	- 0.8
	- 0.9
	- 1.0
	corrupt_seed: 8
	method:
	iNMF:
	LIGER:
	Harmony:
	bindSC:
	CCA_anchor:
	GLUE:
	dim: 50
	alt_dim: 100
	hidden_depth: 2
	hidden_dim: 256
	dropout: 0.2
	lam_graph: 0.02
	lam_align: 0.05
	neg_samples: 10
	seed: 1
	prior_alt:
	data:
	prior:
	gene_region:
	- combined
	- promoter
	extend_range:
	- 0
	- 150000
	corrupt_rate: 0.0
	corrupt_seed: 0
	method:
	iNMF:
	LIGER:
	Harmony:
	bindSC:
	CCA_anchor:
	GLUE:
	dim: 50
	alt_dim: 100
	hidden_depth: 2
	hidden_dim: 256
	dropout: 0.2
	lam_graph: 0.02
	lam_align: 0.05
	neg_samples: 10
	seed: 8
	#------------------------------- Subsample data --------------------------------
	subsample:
	data:
	subsample_size:
	- 250
	- 500
	- 1000
	- 2000
	- 4000
	- 8000
	subsample_seed: 8
	prior:
	gene_region: combined
	extend_range: 0
	corrupt_rate: 0.0
	corrupt_seed: 0
	method:
	UnionCom:
	Pamona:
	MMD_MA:
	iNMF:
	iNMF_FiG:
	LIGER:
	LIGER_FiG:
	Harmony:
	bindSC:
	CCA_anchor:
	GLUE:
	dim: 50
	alt_dim: 100
	hidden_depth: 2
	hidden_dim: 256
	dropout: 0.2
	lam_graph: 0.02
	lam_align: 0.05
	neg_samples: 10
	seed: 1
	#----------------------------- Feature consistency -----------------------------
	feature:
	data:
	prior:
	gene_region: combined
	extend_range: 0
	corrupt_rate: 0.0
	corrupt_seed: 0
	method:
	GLUE:
	dim: 50
	alt_dim: 100
	hidden_depth: 2
	hidden_dim: 256
	dropout: 0.2
	lam_graph: 0.02
	lam_align: 0.05
	neg_samples: 10
	seed: 8

), and then removing all but 2-3 methods in that benchmark.

[Khalid-Usman]

@Jeff1995 , I comment out two datasets and all the methods except GLUE, UnionCom, and Harmony. Still it showing me the above attached error.

[Jeff1995]

Oh that should be a problem with the snakemake profile. Unless you are using a slurm cluster, the local profile should be used:

snakemake -j32 --profile profiles/local  # On local server

[Khalid-Usman]

@Jeff1995 I have modified the attached two files, please let me know if I am supposed to change other files, like GLUE -> evaluation -> workflow -> methods.smk or another file to run the code successfully.

config.docx
display.docx

[Khalid-Usman]

@Jeff1995 When I run the following command after the above configurations. I got the error as attached.

[Khalid-Usman]

Then, I tried to perform some experiments and got the attached error. I am confused about what I am doing wrong.

[Jeff1995]

Sorry for the late reply! Could you check the log files for more detailed errors? E.g., "s01_preprocessing_nbconvert.log"?