krishnanlab / obnb Goto Github PK

• #13
• #14
• #15
• #16
• #17
• #18
• #19
• #20
• #21
• #22
• #23
• #24
• #25
• #26
• #27
• #28
• #29
• #30
• #31
• #32
• #33
• #34
• #35
• #36
• #37
• #38
• #39
• #41
• #40
• #42
• #43
• #44
• #45
• #46
• #47
• #48
• #50

https://github.com/krishnanlab/NetworkLearningEval/blob/2d9d90d2af61cdabff97b0fb6267e69f1f3f3b1c/src/NLEval/label/LabelsetCollection.py#L187

• src/NLEval/graph/sparse.py:1:1: D100 Missing docstring in public module
• src/NLEval/graph/ontology.py:1:1: D100 Missing docstring in public module
• src/NLEval/graph/base.py:1:1: D100 Missing docstring in public module
• src/NLEval/graph/dense.py:1:1: D100 Missing docstring in public module
• src/NLEval/graph/featurevec.py:1:1: D100 Missing docstring in public module
• src/NLEval/model_trainer/supervised_learning.py:1:1: D100 Missing docstring in public module
• src/NLEval/model_trainer/graphgym.py:1:1: D100 Missing docstring in public module
• src/NLEval/model_trainer/gnn.py:1:1: D100 Missing docstring in public module
• src/NLEval/model_trainer/base.py:1:1: D100 Missing docstring in public module
• src/NLEval/model_trainer/label_propagation.py:1:1: D100 Missing docstring in public module
• src/NLEval/label/collection.py:1:1: D100 Missing docstring in public module
• src/NLEval/label/filters/negative_generator.py:1:1: D100 Missing docstring in public module
• src/NLEval/label/filters/existence_filter.py:1:1: D100 Missing docstring in public module
• src/NLEval/label/filters/range_filter.py:1:1: D100 Missing docstring in public module
• src/NLEval/label/filters/base.py:1:1: D100 Missing docstring in public module
• src/NLEval/label/filters/pairwise_filter.py:1:1: D100 Missing docstring in public module
• src/NLEval/label/filters/value_filter.py:1:1: D100 Missing docstring in public module
• src/NLEval/label/split/partition.py:1:1: D100 Missing docstring in public module
• src/NLEval/label/split/base.py:1:1: D100 Missing docstring in public module
• src/NLEval/label/split/holdout.py:1:1: D100 Missing docstring in public module
• src/NLEval/model/label_propagation.py:1:1: D100 Missing docstring in public module
• src/NLEval/data/disgenet.py:1:1: D100 Missing docstring in public module
• src/NLEval/data/go.py:1:1: D100 Missing docstring in public module
• src/NLEval/data/hippie.py:1:1: D100 Missing docstring in public module
• src/NLEval/data/humannet.py:1:1: D100 Missing docstring in public module
• src/NLEval/data/base.py:1:1: D100 Missing docstring in public module
• src/NLEval/data/biogrid.py:1:1: D100 Missing docstring in public module
• src/NLEval/data/bioplex.py:1:1: D100 Missing docstring in public module
• src/NLEval/data/funcoup.py:1:1: D100 Missing docstring in public module
• src/NLEval/data/string.py:1:1: D100 Missing docstring in public module

https://github.com/krishnanlab/NetworkLearningEval/blob/c16104a9a3020c355775523b38526118b4a7f1c4/src/NLEval/valsplit/Holdout.py#L93-L94

According to the definition of reverse, if reverse is true, then should train on bot, i.e., train_ID_ary should actually be set to bot_list instead of top_list.

https://github.com/krishnanlab/NetworkLearningEval/blob/c16104a9a3020c355775523b38526118b4a7f1c4/src/NLEval/valsplit/Base.py#L93

• Check to see if there is a unit test associated with it, if so need to correct it test criterion, otherwise provide a test and confirm the above.

https://github.com/krishnanlab/NetworkLearningEval/blob/c16104a9a3020c355775523b38526118b4a7f1c4/src/NLEval/valsplit/Base.py#L55-L121

Create a new class for train-val-test split (maybe inherit some part from BaseHoldout). Need to automatically perform hyperparameter tuning based on validation scores, taken as in put the objective scoring function, via grid search.

https://github.com/krishnanlab/NetworkLearningEval/blob/1a75b5b1e4525dce81e261d33d13c151cf595135/src/NLEval/label/LabelsetCollection.py#L344-L351

There is a potential issue with the current scheme that would lead to unnecessary removal of label sets. For example, in an iteration, multiple labelsets are to be removed by the LabelsetRangeFilterTrainTestPos filter, but some of them might be okay after reassigning the train/val/test splits.

One potential solution would be reassigning train/val/test splits after every removal of a labelset. However, this poses potential issues such as slower runtime, and also the ordering of removal, which might also lead to different final solution.

When I tried to run sh get_data.sh or bash get_data,sh on my Macbook I got the following error

get_data.sh: line 3: realpath: command not found
usage: dirname path

Add a method copy() analogous to that from list tin BaseGraph o return a deep copy of the graph

• src/NLEval/graph/sparse.py
• src/NLEval/graph/ontology.py
• src/NLEval/graph/base.py
• src/NLEval/graph/featurevec.py
• src/NLEval/model_trainer/base.py
• src/NLEval/label/collection.py
• src/NLEval/label/filters/negative_generator.py
• src/NLEval/label/filters/existence_filter.py
• src/NLEval/label/filters/range_filter.py
• src/NLEval/label/filters/base.py
• src/NLEval/label/filters/pairwise_filter.pyd
• src/NLEval/label/filters/value_filter.py
• src/NLEval/label/split/base.py
• src/NLEval/data/disgenet.py
• src/NLEval/data/go.py
• src/NLEval/data/base.py

https://github.com/krishnanlab/NetworkLearningEval/blob/d394cf357a5b401d828fcf9038943a0c22afb137/src/NLEval/wrapper/ParWrap.py#L6

https://github.com/joblib/joblib

https://github.com/krishnanlab/NetworkLearningEval/blob/d394cf357a5b401d828fcf9038943a0c22afb137/src/NLEval/wrapper/ParWrap.py#L6

In the current setup, the order of the output is (more or less) random. To align back to the original task order, effort must be made to report the task index in the worker function before feeding in to ParDat, which is then used to decide where the result should be placed in the final result array/list. It might be favorable to have a version of ParDat that does the aligning automatically and return the ordered result list.