Uniquifier.py

class UniqueFilter:

    """

    Class to get unique GTF-results, filtered by specified cell-/tissuetypes

    """

    def __init__(self, ense, ucsc, org_filter=None):
        self.results = self.get_filtered_results(org_filter, ense, ucsc)

    def get_results(self):
        return self.results

    def get_filtered_results(self, org_filter, ense, ucsc):

        # Apply Filter

        unfiltered_results = self.concat_without_duplicates(ense, ucsc)
        if org_filter:
            filterstrings = [x+">ACTIVE" for x in org_filter]
            return_list = []
            for element in unfiltered_results:
                if any(tissue in element[-1] for tissue in filterstrings):
                    return_list.append(element)
            return return_list
        else:
            return unfiltered_results


    @staticmethod
    def concat_without_duplicates(ense, ucsc):

        # Concat UCSC and Ensembl data without duplicates

        results = ense+ucsc
        for ens in ense:
            for uc in ucsc:
                if ens[0] == uc[0]:
                    #print("Chromosome Equal")
                    if ense[3] == uc[3]:
                        results.remove(ucsc)

        return results