Stuck in SVD-EM #28
Description
In some cases with missing assays with samples, it seems that SUMO get stuck in the svdEM routine. Here is the output from one such run of sumo run
#Loading file: cnprofile.tsv.gz
Number of dropped rows/features: 0
Number of dropped columns/samples: 0
Data shape: (6756, 344)
#Loading file: gene_expression.tsv.gz
Number of dropped rows/features: 0
Number of dropped columns/samples: 0
Data shape: (2712, 300)
#Loading file: methylation.tsv.gz
Number of dropped rows/features: 18
Number of dropped columns/samples: 0
Data shape: (5361, 337)
#Setting all 'missing' parameters to 0.1
#Total number of unique samples: 345
#Layer: 0
Feature matrix: (345 samples x 6756 features)
Adjacency matrix (345, 345) created [similarity method: euclidean]
Adjacency matrix plot saved to expr_meth_cn/adjacency_0.png
#Layer: 1
Feature matrix: (345 samples x 2712 features)
Adjacency matrix (345, 345) created [similarity method: euclidean]
Adjacency matrix plot saved to expr_meth_cn/adjacency_1.png
#Layer: 2
Feature matrix: (345 samples x 5361 features)
Adjacency matrix (345, 345) created [similarity method: euclidean]
Adjacency matrix plot saved to expr_meth_cn/adjacency_2.png
#Output file expr_meth_cn/prepare.npz created
Directory 'expr_meth_cn/results' already exist and will be overwritten.
Number of clusters ('k'): [2]
Sparsity values ('eta'): [0.1]
#Loading file: expr_meth_cn/prepare.npz
#Number of found graph layers: 3
#Loading sample labels file: labels.tsv
#Found 2 unique sample labels for 90/345 samples
#Number of samples randomly removed in each run: 17 out of 345
#SVD-EM for imputation of missing values in average adjacency matrix
- Initialized 8 missing sample-sample similarities (~0%)
- Iteration(0): Relative change in trace norm: 1.4764676362149612e+18
- Iteration(1): Relative change in trace norm: 0.0
SUMO gets stuck here and does not seem to proceed.