Error balance_DS function

[PrestonLeung]

Hi!

Dolphyn is awesome! I found the github page via the publication!

I just wanted to ask what Python Version do you use for running Dolphyn?

I'm running python 3.12 and I get an error because random package (I think past python 3.11?) do not take in sets when selecting random samples. (similar to something described here benedekrozemberczki/littleballoffur#27)

I was running through the jupyter tutorial to learn how to use Dolphyn, but when I get to:
ge = D.findEpitopes(testrun = 2, protein_seq_file = protein_seq_file, epitile_size=15, epitope_probability_cutoff = 0.5)

I get errors about sets not supported in random.sample()

I fiddled a bit with the function:

def balance_DS(X, y, random_state=42):
    wildtypeIDs = set([item[0] for item in X.index.str.split("_")])
    random.seed(random_state)
    size_smaller_group = y.value_counts().min()
    pos_IDs = set(y[y==1].index)
    # pos_IDs = list(set(y[y==1].index)) # I just made sets into lists
    
    neg_IDs = set(y[y==0].index)
    # neg_IDs = list(set(y[y==0].index)) # here

    neg_IDs= random.sample(neg_IDs, size_smaller_group)
    pos_IDs= random.sample(pos_IDs, size_smaller_group)

    balancedIDs = set(pos_IDs + neg_IDs)
    # balancedIDs = list(set(pos_IDs + neg_IDs)) # and here

    y_bal = y.loc[balancedIDs,]
    X_bal = X.loc[balancedIDs,]    
    return(X_bal, y_bal)

This seemed to fix my errors. Do you also see this?

Cheers and thanks!

[YihuiSun]

It is because of the version of pandas. The older version is working well. What I'm using is pandas_1.3.3.