Add Seurat/MULTI-seq #6168
Add Seurat/MULTI-seq #6168
Conversation
|
There was a problem hiding this comment.
Mostly recommend some tidying up.
I always worry a little about smushing lots of custom stuff (the plots here) into a module that should really be as thin-as-possible a wrapper around some underlying function. But might be a me thing.
Also just wanted to flag the efforts that have gone on to build CLI parts for Seurat, in case you're interested.
|
|
||
| # All values from ext.args are stored as strings | ||
| # Function to transform strings to the correct class | ||
| convert_element <- function(x) { |
There was a problem hiding this comment.
Could you move the functions to the top, with the other function? Would help with readability.
Also, add the proper roxygen-style function documentation to all.
There was a problem hiding this comment.
I totally agree with you on having a wrapper as thin as possible, I'll check if some parts of the CLI can be applied here, I also thought from the beginning that the block of code for the plots could be too large for the module, however splitting the code would imply adding an "intermediate layer" in which we would obtain the results from demultiplexing as an RDS object and produce plots, the problem is that not all plots included in the tool are available under the CLI
merged code from master
| then { | ||
| assertAll( | ||
| { assert process.success }, | ||
| { assert path(process.out.assignment.get(0).get(1)).exists() }, |
There was a problem hiding this comment.
Is there nothing stable in this file to check other than its existence?
There was a problem hiding this comment.
It is a csv file that contains different values for each sample, I tried with { assert path(process.out.classification.get(0).get(1)).exists() } but any idea of another test that might be useful here?
| assay = "HTO" | ||
| nfeatures = 2000 | ||
| quantile = 0.7 | ||
| autoThresh = false | ||
| maxiter = 5 | ||
| qrange_from = 0.1 | ||
| qrange_to = 0.9 | ||
| qrange_by = 0.05 | ||
| verbose = true | ||
| selection_method = "mean.var.plot" | ||
| normalization_method = "CLR" | ||
|
|
||
| // Parameters to generate plots | ||
| group_cells_feature_scatter = "MULTI_ID" | ||
| feature_scatter_feature_1 = "MS-11" | ||
| feature_scatter_feature_2 = "MS-12" | ||
| number_of_features_ridge_plot = 2 | ||
| number_of_cols_ridge_plot = 2 | ||
| group_cells_violin_plot = "MULTI_classification" | ||
| features_violin_plot = "nCount_RNA" | ||
| pt_size = 0.1 | ||
| log = true | ||
| subset_idents = "Negative" | ||
| subset_invert = true | ||
| tsne_scale_data_verbose = false | ||
| run_pca_approx = false | ||
| run_tsne_dim_max = 2 | ||
| run_tsne_perplexity = 100 | ||
| check_duplicates_tsne = false | ||
| resolution = 0.6 | ||
| singlet_identities_tsne = "MULTI_classification" |
There was a problem hiding this comment.
Are these all optional? Does it work without this?
There was a problem hiding this comment.
For all the parameters before // Parameters to generate plots, the tool has the values by default.
After // Parameters to generate plots, some names such as feature_scatter_feature_1 = "MS-11" depend on the data and, therefore, must be given, if the option produce_plots is true, should I include those parameters as inputs in the main.nf?
PR checklist
Closes #XXX
versions.ymlfile.labelnf-core modules test <MODULE> --profile dockernf-core modules test <MODULE> --profile singularitynf-core modules test <MODULE> --profile condanf-core subworkflows test <SUBWORKFLOW> --profile dockernf-core subworkflows test <SUBWORKFLOW> --profile singularitynf-core subworkflows test <SUBWORKFLOW> --profile conda