utils/preprocess.py

import pandas as pd


def normalize_counts(counts: pd.DataFrame):

    if "Patient" in counts.columns:
        counts = counts.set_index("Patient")

    counts = counts.apply(
        pd.to_numeric,
        errors="coerce"
    )

    counts = counts.fillna(0)

    counts_sum = counts.sum(axis=1)

    counts_sum = counts_sum.replace(0,1)

    cpm = counts.div(
        counts_sum,
        axis=0
    ) * 1e6

    return cpm