Index correspondence problem of mutation effect prediction

[Lsxlsxls]

Hello，I have a question in Mutation Effect Prediction.ipynb. When predicting the effects of mutation for a single mutation in PABP, for example [(126,"G","A")], like this:

pabp_data_helper.delta_elbo(pabp_vae_model,[(126,"G","A")], N_pred_iterations=500)

In the function delta_elbo, it calls the dictionary uniprot_focus_col_to_focus_idx, seems like it has wrong corresponding realtions between focus_col and focus_index. I think one to one correspondence between focus_col and focus_seq_trimmed_index is needed because the focus sequence has removed lowercase letters. Otherwise, the position of 126 G to A will mistakenly become the position of ’134‘ H to A. I don't know if I'm right. Please have a look. Thank you.