Command line arguments for target data clustering - nicococo/scRNA GitHub Wiki

Target Dataset Clustering

scRNA-target.sh

Input and output files:

Command line arguments	Description
--src-fname	Source *.npz result filename from Step 2
--fname	Target data (TSV file)
--fgene-ids	Target gene ids (TSV file)
--fout	Result files will use this prefix
--flabels	(optional) Target cluster labels (TSV file)

Data pre-processing Gene/cell filtering arguments (SC3 inspired):

Command line arguments	Description
--min_expr_genes	(Cell filter) Minimum number of expressed genes (default 2000)", default=2000, type=int)
--non_zero_threshold	(Cell/gene filter) Threshold for zero expression per gene (default 1.0)
--perc_consensus_genes	(Gene filter) Filter genes that coincide across a percentage of cells (default 0.98)
--no-cell-filter	Disable cell filter
--no-gene-filter	Disable gene filter
--no-transform	Disable log2(x+1) data transformation

SC3-specific distances and transformations:

Command line arguments	Description
--sc3-dists	(SC3) Comma-separated MTL distances (default euclidean)
--sc3-transf	(SC3) Comma-separated transformations (default pca)

Test settings: The software will cluster any combination of the cluster-range and mixtures and store results separately.

Command line arguments	Description
--cluster-range	Comma separated list of clusters (default 6,7,8)
--mixtures	Comma separated list of convex combination src-trg mixture coefficient (0.=no transfer, default 0.0,0.1,0.2)

Additional commands:

Command line arguments	Description
--no-tsne	Do not plot t-SNE plots as they can be quite time consuming