reymond-group / mlpeptide Goto Github PK

Dear Alice and reymond-group,

thanks for very much for your share of codes.
I am trying to reproduce your codes, in the start of '09-Property-analysis.ipynb',
df_all_nohem = pd.read_pickle(folder +"/pickles/all_sequences_with_NN_prop_helicity.pkl")
But I checked all your 13 ipynb, there is not another "all_sequences_with_NN_prop_helicity.pkl" avaible.
Could you review your codes to check this issues?
thanks a lot in advance!

Rocks Dong

Hi Alice and reymond-group,

thanks very much for your paper "Machine learning designs non-hemolyticantimicrobial peptides" and shared codes!
Your awesome isolutions is a ensemable of simple RNN models to solve a nice issue!
I am trying to reproduce your codes and may hope replace GRU with LSTM in future.
In 01b-hemolysis-data-extraction.ipynb, the codes call dbassp api link, "https://dbaasp.org/api/v1", but the link is invalid.
I goto dbassp website, the current api version is v3 and need pre apply a key from dbaasp.
Currently, I hope firstly to reproduce your codes.
Could you share the "DAASP_RNN_dataset_with_hemolysis" data in .csv or .plk(pandas pickle) which is created in 01b step?

Later, we will apply the key for dbaasp api v3 after I can run your codes throughtly.
thanks a lot in advance!!
BTW, why not use LSTM but use GRU? I guess it's because the data volume is small and so GRU is enough?

Rocke dong(Dong qichang)