Preprocessing benchmarking #337
Conversation
There was a problem hiding this comment.
Hi @xingzhongyu, thanks for the PR! Could you provide a bit more context about why this edit (targs[1]) is necessary? From my understanding, targs is the of gene indexs to hold out, and targs[1] would be the second gene index to holdout. So I'm a bit confused about the reason for this change
There was a problem hiding this comment.
Because the loop uses enumerate, the resulting tuple needs to be taken out of the following content, otherwise report an error
There was a problem hiding this comment.
@xingzhongyu could you help make some preliminary efforts to migrate the current download functions for the imputation dataset to use the downloading utilities in dance/utils/download.py like in the clustering dataset class? Imputation dataset is currently the only dataset that uses wget and curl (for historical reasons that are no longer relevant now).
There was a problem hiding this comment.
OK I will finish the task in this week
for more information, see https://pre-commit.ci
for more information, see https://pre-commit.ci
for more information, see https://pre-commit.ci
Added modal prediction dataset;
Limit the number of features extracted in construct_enhanced_feature_graph in dance/transforms/graph/scmogcn_graph.py;
The extraction of idx in the get_feature method in dance/data/base.py is restricted to provide slices within the number of features.