yhr91/gears_misc

gears_misc's Issues

Please upload data

Hi, thanks for sharing the reproducible notebooks for the paper. Could you please upload the data used in the notebooks to make them actually reproducible?

Publication of model parameters

Hi!

I have posted an issue in the snap-stanford github repo of the final GEARS model. Since I didn't know which repo is more likely to catch your attention, I simply want to link it here: snap-stanford/GEARS#50 . Feel free to answer in any one of the two repos.

Thank you for your time.

Reproduce figure 2 b-g

Hi, could you provide code to reproduce figure 2b-g in your preprint? Or at least could tell how to rerun the analysis workflow?

Reproduce figure 3

Hi, I am trying to reproduce the analysis in figure 3, particularly the UMAPs. Could you let me know how to do it?

Some related questions:

1. Should I use the script https://github.com/yhr91/gears_misc/blob/main/paper/Fig3_UMAP_predict.py and https://github.com/yhr91/gears_misc/blob/main/paper/Fig3_UMAP_train.py?

2. It seems the Fig3_UMAP_train.py only saved the model, so how did you generate the embeddings used in figure 3b?

3. Similarly, what embedding vectors you used to generate figure 3c? Is it related to this line https://github.com/yhr91/gears_misc/blob/73a026049d22107ef1a31c81cb677f646b62dea1/paper/Fig3_UMAP_predict.py#L69

Thank you!

Missing data file when trying to reproduce results

Hi Yusuf and Kexin!

Thanks for developing the great tool! I'm trying to reproduce the result in Fig. 2f of the manuscript. According to the instructions, I'm trying to run the paper/fig2_train.py script, but the following error was thrown:

Traceback (most recent call last):
  File "fig2_train.py", line 63, in <module>
    gears_model.model_initialize(hidden_size = 64, no_perturb = no_perturb, go_path = go_path)
  File "<REDACTED>/GEARS_misc/paper/../gears/gears.py", line 147, in model_initialize
    edge_list = get_similarity_network(network_type = 'go', adata = self.adata, threshold = coexpress_threshold, k = num_similar_genes_go_graph, gene_list = self.pert_list, data_path = self.data_path, data_name = self.dataset_name, split = self.split, seed = self.seed, train_gene_set_size = self.train_gene_set_size,
set2conditions = self.set2conditions, gi_go = self.gi_go, dataset = go_path)
  File "<REDACTED>/GEARS_misc/paper/../gears/utils.py", line 169, in get_similarity_network
    df_jaccard = pd.read_csv('/dfs/user/kexinh/gears2/go_essential_all.csv')
  File "<REDACTED>/GEARS_misc/paper/conda/lib/python3.8/site-packages/pandas/io/parsers/readers.py", line 912, in read_csv
    return _read(filepath_or_buffer, kwds)
  File "<REDACTED>/GEARS_misc/paper/conda/lib/python3.8/site-packages/pandas/io/parsers/readers.py", line 577, in _read
    parser = TextFileReader(filepath_or_buffer, **kwds)
  File "<REDACTED>/GEARS_misc/paper/conda/lib/python3.8/site-packages/pandas/io/parsers/readers.py", line 1407, in __init__
    self._engine = self._make_engine(f, self.engine)
  File "<REDACTED>/GEARS_misc/paper/conda/lib/python3.8/site-packages/pandas/io/parsers/readers.py", line 1661, in _make_engine
    self.handles = get_handle(
  File "<REDACTED>/GEARS_misc/paper/conda/lib/python3.8/site-packages/pandas/io/common.py", line 859, in get_handle
    handle = open(
FileNotFoundError: [Errno 2] No such file or directory: '/dfs/user/kexinh/gears2/go_essential_all.csv'

This I believe is the line where the error was raised. It seems that there are some hard-coded file paths that have not been provided in either yhr91/GEARS_misc or snap-stanford/GEARS. Could you help point me to where I can obtain these files so that I might be able to reproduce the result in the paper?

Best!
Zhi-Jie

Missing files for reproducing SI Fig 17 Horlbeck Jurkat

Hi Yusuf and Kexin,

I'm trying to reproduce SI Figure 17 of predictions for the Horlbeck dataset from this notebook: https://github.com/yhr91/GEARS_misc/blob/main/paper/SI_Fig_17_Horlbeck_Jurkat.ipynb

So far I've encountered a couple of issues.

1. The notebook expects a dataset h5ad file for the dataset name horlbeck_gi_jurkat.
2. The path for /dfs/user/kexinh/gears2/data/pert_genes_gi.pkl is hard-coded into pertdata.py.

Neither file is included in the repo, nor can I locate the files from the main GEARS repository. Could you help point me to where I could access these files?

Thank you!