def qc_charr_score(self, df: pd.DataFrame) -> float:
        """
        calculate charr score
        Args:
            df(pd.DataFrame): dataframe of parser vcf
        Returns:
            float, charr score
        """

        def charr_score(row: pd.Series) -> float:
            """
            charr score for one site
            Args:
                row(pd.Series): one site
            Returns:
                float, charr score
            """
            ad = [int(i) for i in row["ad"].split(",")]  # allele depth
            freq = float(row["freq"])
            if (ad[0] + ad[1]) == 0 or freq == 0:
                return 0  # The denominator cannot be zero
            return ad[0] / (freq * (ad[0] + ad[1]))

        # autosome
        data = df[df["chrom"].isin([f"chr{i}" for i in range(1, 23)])]

        data = data[data["gt"].isin(["1/1", "1|1"])]  # Homogenous variant sites
        data = data[(data["dp"] > 30) & (data["gq"] > 20)]  # filter

        def get_freq(row: pd.Series) -> float:
            """
            get reference freq for one site in population
            Args:
                row(pd.Series): one site
            Returns:
                float, reference freq
            """
            key = (row["chrom"], row["position"])
            return self.charr_required_freq.get(key, 0.0001)  # If not, fill in the minimum value to avoid division by 0

        data["freq"] = data.apply(get_freq, axis=1)
        if data.shape[0] == 0:
            return 0.0  # no site after filter
        score = round(data.apply(charr_score, axis=1).sum() / data.shape[0], 4)  # mean and retain four decimal places

        return score