Could you provide split files for different experiments on Norman

[igor-sadalski]

it seems that unseen split has no seed set before shuffling, specifically the unseen split missing seed is at data.py:162:

                                                                                                                                                                                                                     
  155:                    for i in range(5):                                                                                                                                                                                  
  156:                        perturbations = np.unique(self.adata.obs['condition'])                                                                                                                                        
  157:                        double_perturbation = [p for p in perturbations if 'ctrl' not in p]                                                                                                                             
  158:                        single = []                                                                                                                                                                                     
  159:                        [single.extend(p.split('+')) for p in double_perturbation]
  160:                        single = list(set(single))
  161:
  162:                        shuffle(single)   # <-- no seed set before this
  163:                        remove_genes = single[:12]

[oYCLo]

Thank you for your interest in our work! We have uploaded the new data for download along with the corresponding split data here: https://drive.google.com/drive/folders/1cNpYAt9jVWZN82miNZtkP10YeSo7hufL. We have also updated the README file accordingly. We hope this helps resolve your issue. Please do not hesitate to contact me if you have any further questions.