Open jmche opened 1 year ago
Hi @jmche, thank you for your interest in this project. You are correct that the expected input of the model is Level 3 L1000 data. Since all training, validation, and predictions were made using Level 3 data, we unfortunately do not recommend transforming Level 5 signatures with our model, as the quality of the results are unknown.
I download the pre-trained model from https://maayanlab.cloud/sigcom-lincs/#/Download I found L1000_to_RNAseq_pipeline_training.ipynb said the input is Level 3 of L1000.
I used the DeepCE model to predict a new drug and would get the level 5 of L1000 since its input is Level 5. I want to know if the input of the pre-trained model was used the Level 3 of L1000. Is that mean I couldn't use the pre-trained model for my output(Level 5 of L1000) to transfer to RNA-seq profiles?