predict_single_sample_DLBCLone#28
Conversation
| select(sample_id,V1,V2) %>% | ||
| column_to_rownames("sample_id") | ||
|
|
||
| train_labels = dplyr::filter( |
There was a problem hiding this comment.
By doing this instead of using train_projection$df you are trusting that the row order has not changed, which may not be true.
There was a problem hiding this comment.
I see, Ill fix this. Though I thought the lymphgen coulmn are truth labels that wouldnt change as they are linked to the sample_id and not V1 and V2
There was a problem hiding this comment.
Yes but the function may drop some rows so this would either cause a mismatch between sample and label or probably trigger an error if the lengths ended up different.
| conf_threshold = best_params$threshold, | ||
| na_label = "Other", | ||
| use_weights = best_params$use_w, | ||
| ignore_top = FALSE |
There was a problem hiding this comment.
I think this can't be hard-coded. We should discuss
|
|
||
| if(drop_unlabeled_from_training){ | ||
| pred = dplyr::filter(pred,sample_id %in% test_df$sample_id) | ||
| } |
There was a problem hiding this comment.
What is the point of this?
There was a problem hiding this comment.
if(drop_unlabeled_from_training){
pred = dplyr::filter(pred,sample_id %in% test_df$sample_id)
}
is now removed. I kept it from the original function predict_single_sample() as it appeared it had some sort of sanity checking functionality but now that you point it out it doesnt appear to have a purpose!
rdmorin
left a comment
rdmorin
left a comment
There was a problem hiding this comment.
Great start. Just a few comments and at least one change depending on how you respond to the comments
…ata on optimized UMAP. predictions_df param is redundant as a result
…test' labeling issue
…lot and param documentation
…elf, fully incorporated optimized_model to predict_single_sample
…dict_single_sample function
…n .rds for reproducibility, , , and also added to DLBCLone_save_optimized for reproducibility + running without error when running predict_single_sample
…w be used as test_df for training puposes reproducibly
…e_sample list output objects and adding necessary objects
…t to accept DLBCLone_KNN and DLBCLone_optimize_params list outputs
|
Is there anything in this PR that hasn't already made it into the Master branch? If not, @lklossok , can you close this PR? |
2 participants
No description provided.