Merge pull request #7 from mkandziora/dev

Dev

PhylUp/phyl_up.py

@@ -53,7 +53,9 @@ def __init__(self, id_to_spn, aln, aln_schema, tre, tre_schema, config, ignore_a
         self.aln = DnaCharacterMatrix.get(path=self.aln_fn, schema=self.aln_schema)
         self.tre_fn = tre
         self.tre_schema = tre_schema
-        self.tre = None
+        if not self.tre_fn == None:
+            self.tre = Tree.get(path=os.path.abspath(self.tre_fn), schema=self.tre_schema,
+                                    taxon_namespace=self.aln.taxon_namespace, preserve_underscores=True)
         self.ignore_acc_list = ignore_acc_list
         sys.stdout.write('Translate input names to ncbi taxonomy...\n')
         self.table = phylogenetic_helpers.build_table_from_file(id_to_spn, self.config, self.config.downtorank)
@@ -850,7 +852,8 @@ def prefer_taxa_from_locus(self, ns_txid_df):
                     for tax_id in preferred_taxa_ids:
                         # if tax_id_from_df is lower taxon from preferred ids:
                         part_tax_id_from_df = self.ncbi_parser.match_id_to_mrca(tax_id_from_df, tax_id)
-                        if part_tax_id_from_df != 0:
+                        # if part_tax_id_from_df != 0:
+                        if part_tax_id_from_df is True:
                             new_filtered_taxid_df = ns_txid_df
                         else:
                             new_filtered_taxid_df = ns_txid_df[0:0]
@@ -1156,7 +1159,8 @@ def filter(self, new_seqs):
             if type(self.mrca) == int:
                 # TODO: make self.mrca to be a set - single id is type int
                 print("DO I EVER GET HERE - MRCA IS INT")
-                if mrca_tx == self.mrca:
+                #if mrca_tx == self.mrca:
+                if mrca_tx == True:
                     to_add = new_seqs[select_tf]
                     assert len(to_add) != 0, len(to_add)
                     self.upd_new_seqs = pd.concat([self.upd_new_seqs, to_add], axis=0, ignore_index=True, sort=True)
@@ -1166,8 +1170,9 @@ def filter(self, new_seqs):
                     to_del.at[:, 'status'] = 'deleted - mrca'
                     self.del_table = pd.concat([self.del_table, to_del], axis=0, ignore_index=True, sort=True)
             elif type(self.mrca) == set:
-                # print(mrca_tx in self.mrca, mrca_tx, self.mrca)
-                if mrca_tx in self.mrca:
+                #print(mrca_tx in self.mrca, mrca_tx, self.mrca)
+                # if mrca_tx in self.mrca:
+                if mrca_tx == True:
                     to_add = new_seqs[select_tf]
                     assert len(to_add) != 0, len(to_add)
                     self.upd_new_seqs = pd.concat([self.upd_new_seqs, to_add], axis=0, ignore_index=True, sort=True)
@@ -1227,7 +1232,12 @@ def __init__(self, config, table):
     def filter(self, new_seqs):
         debug('FilterUniqueAcc')
         # delete things in table
-        new_seqs_unique = new_seqs.drop_duplicates(subset=['accession'], keep='first')  #todo: keep longest - might affect length filter
+        new_seqs_unique = drop_shortest_among_duplicates(new_seqs)
+        #new_seqs_unique_old = new_seqs.drop_duplicates(subset=['accession'], keep='first')  #todo: keep longest - might affect length filter
+        #assert new_seqs_unique_old['sseq'].isin(new_seqs_unique['sseq']).all()
+        #assert new_seqs_unique_old['accession'].isin(new_seqs_unique['accession']).all()
+        #assert new_seqs_unique.sseq == new_seqs_unique_new.sseq
+
         drop = new_seqs.drop(new_seqs_unique.index.tolist())
 
         self.upd_new_seqs = new_seqs_unique
@@ -1240,6 +1250,24 @@ def filter(self, new_seqs):
         write_msg_logfile(msg, self.config.workdir)
 
 
+def drop_shortest_among_duplicates(new_seqs):
+    """
+    Keep the entry with the longest sequences from the duplicated accessions.
+
+    :param new_seqs:
+    :return:
+    """
+    accs = new_seqs.drop_duplicates(subset=['accession'], keep='first')
+
+    new_seqs['seq_len'] = new_seqs['sseq'].apply(len)
+    grouped_df = new_seqs.groupby("accession")
+    maximums_idx = grouped_df.seq_len.idxmax()
+    maximums = new_seqs.loc[maximums_idx]
+
+    assert len(maximums) == len(accs.accession), (len(maximums), len(accs.accession))
+    return maximums
+
+
 def check_df_index_unique(df):
     assert df.index.is_unique == True, df.index
 

install_req.bash

@@ -62,4 +62,8 @@ echo "export PATH=$PWD:\$PATH" >> ~/.bashrctest
 cd ..
 cd ..
 
+wget 'https://ftp.ncbi.nlm.nih.gov/pub/taxonomy/taxdump.tar.gz'
+gunzip  -cd taxdump.tar.gz | (tar xvf - names.dmp nodes.dmp)
+mv *.dmp ./data/
+
 source ~/.bashrc

tests/test_filter_length.py

@@ -78,3 +78,4 @@ def test_filterlen():
 
     assert before > after
     assert del_tab > 0
+    assert before == after + del_tab

tests/test_filter_otu.py

@@ -1,4 +1,5 @@
 import os
+import pandas as pd
 import pytest
 
 from distutils.dir_util import copy_tree
@@ -36,22 +37,58 @@ def configure():
     pytest.test = test
 
     new_seqs = test.extend()
+    assert len(new_seqs) > 0, len(new_seqs)
+
     aln = phylogenetic_helpers.read_in_aln(test.aln_fn, test.aln_schema)
 
     new_seqs = new_seqs[~new_seqs['accession'].isin(test.table['accession'])]  # ~ is the pd not in/!
-    new_seqs = test.basic_filters(aln, test.mrca, new_seqs)
+    assert len(new_seqs) > 0, len(new_seqs)
+
+    #new_seqs = test.basic_filters(aln, test.mrca, new_seqs)
+    orig_len = new_seqs
+    columns = ['ncbi_txn', 'ncbi_txid', 'status', 'status_note', "date",
+               'accession', 'pident', 'evalue', 'bitscore', 'sseq', 'title']
+    all_del = pd.DataFrame(columns=columns)
+
+    remove_basics = [phyl_up.FilterUniqueAcc(test.config, test.table),
+                     # FilterBLASTThreshold(self.config),
+                     phyl_up.FilterLength(test.config, aln),
+                     phyl_up.FilterMRCA(test.config, test.mrca)
+                     ]
+    for f in remove_basics:
+        print(f)
+        internal_new_seq = new_seqs
+        f.filter(new_seqs)
+        new_seqs = f.upd_new_seqs
+        del_seq = f.del_table
+        # all_del = all_del.append(del_seq, ignore_index=True)
+        all_del = pd.concat([all_del, del_seq], ignore_index=True, sort=True)
+        phyl_up.check_filter_numbers(del_seq, new_seqs, internal_new_seq)
+        assert len(new_seqs) > 0, len(new_seqs)
+
+        if len(new_seqs) == 0:
+            return new_seqs  # stop filtering if new seqs is 0
+    phyl_up.check_filter_numbers(all_del, new_seqs, orig_len)
+    assert len(new_seqs) > 0, len(new_seqs)
+
     # next filter need infos in table
     new_seqs = test.add_new_seqs(new_seqs)
     pytest.new_seqs = new_seqs
 
+    assert len(pytest.new_seqs) > 0, len(pytest.new_seqs)
+
 
 def test_filter_otu_no_rank():
     test = pytest.test
+    print(test.table)
     new_seqs = pytest.new_seqs
+    assert len(new_seqs) > 0, len(new_seqs)
+
     conf = pytest.conf
 
 
     before = len(new_seqs)
+    print(before)
     new_seqs_org = new_seqs
     before_table = deepcopy(test.table)
 

tests/test_filter_seqident.py

@@ -214,6 +214,7 @@ def test_filter_compare():
     conf = config.ConfigObj(configfi, workdir, interactive=False)
     conf.threshold = 2
     conf.blast_folder = os.path.abspath("./data/blast_for_tests")
+    conf.identical_seqs = False
 
     if not os.path.exists(workdir):
         os.rename(workdir)
@@ -229,36 +230,62 @@ def test_filter_compare():
 
     test = phyl_up.PhylogeneticUpdater(id_to_spn, seqaln, mattype, trfn, schema_trf, conf)
 
-    new_seqs = test.extend()
+    new_existing_seq = """AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAATCGAAACCTGCATAGCAGAACGACCTGTGAACATGTAAAACAATTGGGTGTTCTAAGTATCGGGCTCTTGTCCGATTCCTAGGATGCCATGTTGACGTGCGTCTTTGGCAAGCCCCTTGGGTGTCTAAGGACGTCACGTCGACGCAACAACAAACCCCCGGCACGGCATGTGCCAAGGAAATATAAACTTAAGAAGGGCTTGTTCCATGCATTGCCGTTCGTGGTGACTGCATTGAAACTTGCTTCTCTATAATTAATAAACGACTCTCGGCAACGGATATCTCGGCTCACGCATCGATGAAGAACGTAGCAAAATGCGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTTTTTGAACGCAAGTTGCGCCCGAAACCTTTTGGTTGAGGGCACGTCTGCCTGGGCGTCACACATCGCGTCGCCCCCATCACACCTCTTGACGGGGATGTTTGAATGGGGACGAAGATTGGTCTCCTGTTCCTAAGGTGCGGTTGGCTGAATTTTGAGTCCTCTTCGACGGACGCACGATTAGTGGTGGTTGACAAGACCTTCTTATCGAGTTGTGTGTTCCAAGAAGTAAGGAATATCTCTTTAACGACCCTAAAGTGTTGTCTCATGACGATGCTTCGACTGC"""
+
+    test.table.loc[1, 'sseq'] = new_existing_seq
+    test.table.loc[1, 'ncbi_txid'] = 123456
+    test.table.loc[1, 'status'] = 0
+
+    assert test.table['sseq'].str.contains(new_existing_seq).any() == True
 
-    new_existing_seq = """TCGAAACCTGCATAGCAGAACGACCTGTGAACATGTAAAACAATTGGGTGTTCTAAGTATCGGGCTCTTGTCCGATTCCTAGGATGCCATGTTGACGTGCGTCTTTGGCAAGCCCCTTGGGTGTCTAAGGACGTCACGTCGACGCAACAACAAACCCCCGGCACGGCATGTGCCAAGGAAATATAAACTTAAGAAGGGCTTGTTCCATGCATTGCCGTTCGTGGTGACTGCATTGAAACTTGCTTCTCTATAATTAATAAACGACTCTCGGCAACGGATATCTCGGCTCACGCATCGATGAAGAACGTAGCAAAATGCGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTTTTTGAACGCAAGTTGCGCCCGAAACCTTTTGGTTGAGGGCACGTCTGCCTGGGCGTCACACATCGCGTCGCCCCCATCACACCTCTTGACGGGGATGTTTGAATGGGGACGAAGATTGGTCTCCTGTTCCTAAGGTGCGGTTGGCTGAATTTTGAGTCCTCTTCGACGGACGCACGATTAGTGGTGGTTGACAAGACCTTCTTATCGAGTTGTGTGTTCCAAGAAGTAAGGAATATCTCTTTAACGACCCTAAAGTGTTGTCTCATGACGATGCTTCGACTGC"""
+    new_seqs = test.extend()
 
     new_seqs = new_seqs[~new_seqs['accession'].isin(test.table['accession'])]  # ~ is the pd not in/!
 
+
     f = phyl_up.FilterUniqueAcc(test.config, test.table)
     f.filter(new_seqs)
     new_seqs = f.upd_new_seqs
     new_seqs.loc[1, 'sseq'] = new_existing_seq
+    new_seqs.loc[1, 'ncbi_txid'] = 123456
     assert new_seqs.loc[1, 'sseq'] == new_existing_seq
-    assert new_seqs['sseq'].str.contains(new_existing_seq).any()
-    assert test.table['sseq'].str.contains(new_existing_seq).any()
+    assert new_seqs['sseq'].str.contains(new_existing_seq).any() == True
+    assert test.table['sseq'].str.contains(new_existing_seq).any() == True
+    new_seqs.loc[1, 'ncbi_txid'] = 123456
+    new_seq_acc = new_seqs.loc[1, 'accession']
+
 
     new_seqs_before = new_seqs
 
     len_table_before = len(test.table)
     new_seqs = test.add_new_seqs(new_seqs)
     assert new_seqs.index.tolist() != new_seqs_before.index.tolist()
-    # assert new_existing_seq in [new_seqs['sseq']]
+    assert new_seqs['sseq'].str.contains(new_existing_seq).any() == True
     len_table_after = len(test.table)
     assert len_table_before < len_table_after, (len_table_before, len_table_after)
 
     assert test.table['sseq'].hasnans == False, test.table['sseq']
 
     assert new_seqs['sseq'].str.contains(new_existing_seq).any() == True
+    assert new_seqs['accession'].str.contains(new_seq_acc).any() == True
+    assert test.table['sseq'].str.contains(new_existing_seq).any() == True
+    print(new_seqs.index[new_seqs['sseq'].str.contains(new_existing_seq).any()])
+
+#     new_seqs = test.compare_filter(new_seqs)
+    f = phyl_up.FilterSeqIdent(test.config, test.table, test.status)
+    before_filter = new_seqs
+    f.filter(new_seqs)
+    test.aln = f.aln
+    new_seqs = f.upd_new_seqs  # assign new_seqs to last upd from filter before
+    test.table = f.table
+    del_seq = f.del_table
+    print(new_seq_acc)
 
-    new_seqs = test.compare_filter(new_seqs)
     print(test.table['sseq'].str.contains(new_existing_seq).any())
     print(new_seqs['sseq'].str.contains(new_existing_seq).any())
+    print(test.table.loc[1, ])
+    assert new_seqs['accession'].str.contains(new_seq_acc).any() == False
+
     assert new_seqs['sseq'].str.contains(new_existing_seq).any() == False
 
     all_avail_data = test.table[test.table['status'].between(0, test.status, inclusive=True)]
@@ -269,8 +296,7 @@ def test_filter_compare():
     print(found.count())
     assert found.count() == 1, found.count()
 
-
-
+# todo check: seems to be testing the same as above but differnt and fails now
 def test_filter_compare_shorter():
     print('test_filter_seqident')
     workdir = "tests/output/test_run_seqident"
@@ -284,11 +310,32 @@ def test_filter_compare_shorter():
     conf = config.ConfigObj(configfi, workdir, interactive=False)
     conf.threshold = 2
     conf.blast_folder = os.path.abspath("./data/blast_for_tests")
+
+    if not os.path.exists(workdir):
+        os.rename(workdir)
+    if not os.path.exists(workdir):
+        os.mkdir(workdir)
+    tmp_folder = os.path.join(workdir, 'tmp')
+    if not os.path.exists(tmp_folder):
+        os.mkdir(tmp_folder)
+    # call(['cp', '-a', 'data/tmp_for_test/', tmp_folder])
+    copy_tree('data/tmp_for_test/', tmp_folder)
+    shutil.copyfile('data/tiny_test_example/updt_aln.fasta', os.path.join(workdir, 'updt_aln.fasta'))
+    shutil.copyfile('data/tiny_test_example/updt_tre.tre', os.path.join(workdir, 'updt_tre.tre'))
+
     test = phyl_up.PhylogeneticUpdater(id_to_spn, seqaln, mattype, trfn, schema_trf, conf)
 
+    new_existing_seq = """AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAATCGAAACCTGCATAGCAGAACGACCTGTGAACATGTAAAACAATTGGGTGTTCTAAGTATCGGGCTCTTGTCCGATTCCTAGGATGCCATGTTGACGTGCGTCTTTGGCAAGCCCCTTGGGTGTCTAAGGACGTCACGTCGACGCAACAACAAACCCCCGGCACGGCATGTGCCAAGGAAATATAAACTTAAGAAGGGCTTGTTCCATGCATTGCCGTTCGTGGTGACTGCATTGAAACTTGCTTCTCTATAATTAATAAACGACTCTCGGCAACGGATATCTCGGCTCACGCATCGATGAAGAACGTAGCAAAATGCGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTTTTTGAACGCAAGTTGCGCCCGAAACCTTTTGGTTGAGGGCACGTCTGCCTGGGCGTCACACATCGCGTCGCCCCCATCACACCTCTTGACGGGGATGTTTGAATGGGGACGAAGATTGGTCTCCTGTTCCTAAGGTGCGGTTGGCTGAATTTTGAGTCCTCTTCGACGGACGCACGATTAGTGGTGGTTGACAAGACCTTCTTATCGAGTTGTGTGTTCCAAGAAGTAAGGAATATCTCTTTAACGACCCTAAAGTGTTGTCTCATGACGATGCTTCGACTGC"""
+
+    test.table.loc[1, 'sseq'] = new_existing_seq
+    test.table.loc[1, 'ncbi_txid'] = 123456
+    test.table.loc[1, 'status'] = 0
+
+    assert test.table['sseq'].str.contains(new_existing_seq).any() == True
+
     new_seqs = test.extend()
 
-    new_existing_seq = """AGAACGACCTGTGAACATGTAAAACAATTGGGTGTTCTAAGTATCGGGCTCTTGTCCGATTCCTAGGATGCCATGTTGACGTGCGTCTTTGGCAAGCCCCTTGGGTGTCTAAGGACGTCACGTCGACGCAACAACAAACCCCCGGCACGGCATGTGCCAAGGAAATATAAACTTAAGAAGGGCTTGTTCCATGCATTGCCGTTCGTGGTGACTGCATTGAAACTTGCTTCTCTATAATTAATAAACGACTCTCGGCAACGGATATCTCGGCTCACGCATCGATGAAGAACGTAGCAAAATGCGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTTTTTGAACGCAAGTTGCGCCCGAAACCTTTTGGTTGAGGGCACGTCTGCCTGGGCGTCACACATCGCGTCGCCCCCATCACACCTCTTGACGGGGATGTTTGAATGGGGACGAAGATTGGTCTCCTGTTCCTAAGGTGCGGTTGGCTGAATTTTGAGTCCTCTTCGACGGACGCACGATTAGTGGTGGTTGACAAGACCTTCTTATCGAGTTGTGTGTTCCAAGAAGTAAGGAATATCTCTTTAACGACCCTAAAGTGTTGTCTCATGACGATGCTTCGACTGC"""
+ #   new_existing_seq = """AGAACGACCTGTGAACATGTAAAACAATTGGGTGTTCTAAGTATCGGGCTCTTGTCCGATTCCTAGGATGCCATGTTGACGTGCGTCTTTGGCAAGCCCCTTGGGTGTCTAAGGACGTCACGTCGACGCAACAACAAACCCCCGGCACGGCATGTGCCAAGGAAATATAAACTTAAGAAGGGCTTGTTCCATGCATTGCCGTTCGTGGTGACTGCATTGAAACTTGCTTCTCTATAATTAATAAACGACTCTCGGCAACGGATATCTCGGCTCACGCATCGATGAAGAACGTAGCAAAATGCGATACTTGGTGTGAATTGCAGAATCCCGTGAACCATCGAGTTTTTGAACGCAAGTTGCGCCCGAAACCTTTTGGTTGAGGGCACGTCTGCCTGGGCGTCACACATCGCGTCGCCCCCATCACACCTCTTGACGGGGATGTTTGAATGGGGACGAAGATTGGTCTCCTGTTCCTAAGGTGCGGTTGGCTGAATTTTGAGTCCTCTTCGACGGACGCACGATTAGTGGTGGTTGACAAGACCTTCTTATCGAGTTGTGTGTTCCAAGAAGTAAGGAATATCTCTTTAACGACCCTAAAGTGTTGTCTCATGACGATGCTTCGACTGC"""
 
     new_seqs = new_seqs[~new_seqs['accession'].isin(test.table['accession'])]  # ~ is the pd not in/!
 
@@ -297,6 +344,11 @@ def test_filter_compare_shorter():
     new_seqs = f.upd_new_seqs
     new_seqs.loc[1, 'sseq'] = new_existing_seq
     assert new_seqs.loc[1, 'sseq'] == new_existing_seq
+    assert new_seqs['sseq'].str.contains(new_existing_seq).any() == True
+    assert test.table['sseq'].str.contains(new_existing_seq).any() == True
+    new_seqs.loc[1, 'ncbi_txid'] = 123456
+    new_seq_acc = new_seqs.loc[1, 'accession']
+    assert new_seqs.loc[1, 'sseq'] == new_existing_seq
     assert new_seqs['sseq'].str.contains(new_existing_seq).any()
     assert test.table['sseq'].str.contains(new_existing_seq).any()
 
@@ -311,10 +363,21 @@ def test_filter_compare_shorter():
 
     assert test.table['sseq'].hasnans == False, test.table['sseq']
 
-    new_seqs = test.compare_filter(new_seqs)
-    print(new_seqs['sseq'])
+    assert new_seqs['sseq'].str.contains(new_existing_seq).any() == True
+
+    # new_seqs = test.compare_filter(new_seqs)
+    f = phyl_up.FilterSeqIdent(test.config, test.table, test.status)
+    before_filter = new_seqs
+    f.filter(new_seqs)
+    test.aln = f.aln
+    new_seqs = f.upd_new_seqs  # assign new_seqs to last upd from filter before
+    test.table = f.table
+    del_seq = f.del_table
+    print(new_seq_acc)
 
-    print(new_existing_seq)
+    print(test.table['sseq'].str.contains(new_existing_seq).any())
+    print(new_seqs['sseq'].str.contains(new_existing_seq).any())
+    assert new_seqs['accession'].str.contains(new_seq_acc).any() == False
     assert new_seqs['sseq'].str.contains(new_existing_seq).any() == False
 
     all_avail_data = test.table[test.table['status'].between(0, test.status, inclusive=True)]

tests/test_filter_unique.py

@@ -1,4 +1,5 @@
 import os
+import pandas as pd
 from distutils.dir_util import copy_tree
 
 from PhylUp import phyl_up, config
@@ -39,3 +40,53 @@ def test_filter_unique():
     assert before > after
     assert del_tab > 0
     assert after + del_tab == before
+
+def test_filters_to_longest():
+    print('test_filter_unique')
+    workdir = "tests/output/test_runs_unique"
+    trfn = "data/tiny_test_example/test.tre"
+    schema_trf = "newick"
+    id_to_spn = "data/tiny_test_example/test_nicespl.csv"
+    seqaln = "data/tiny_test_example/test.fas"
+    mattype = "fasta"
+    configfi = "data/localblast_test.config"
+
+    if not os.path.exists(workdir):
+        os.mkdir(workdir)
+    tmp_folder = os.path.join(workdir, 'tmp')
+    if not os.path.exists(tmp_folder):
+        os.mkdir(tmp_folder)
+    # call(['cp', '-a', 'data/tmp_for_test/', tmp_folder])
+    copy_tree('data/tmp_for_test/', tmp_folder)
+
+    conf = config.ConfigObj(configfi, workdir, interactive=False)
+    conf.threshold = 2
+    conf.blast_folder = os.path.abspath("./data/blast_for_tests")
+    test = phyl_up.PhylogeneticUpdater(id_to_spn, seqaln, mattype, trfn, schema_trf, conf)
+
+    new_seqs = test.extend()
+
+    # make own longest selection
+    accs = new_seqs.drop_duplicates(subset=['accession'], keep='first')
+
+    len_supposed = len(accs)
+
+    select = pd.DataFrame(columns=['ncbi_txn', 'ncbi_txid', 'status', 'status_note', "date", 'accession',
+                                   'pident', 'evalue', 'bitscore', 'sseq', 'title'])
+    for acc in accs.accession:
+        acc_df = new_seqs[new_seqs.accession == acc]
+        lenseq_acc_df = acc_df['sseq'].apply(len)
+        idxmax_val = lenseq_acc_df.idxmax()
+        select = select.append(acc_df.loc[idxmax_val])
+
+    assert len(select) == len_supposed, (len(select), len_supposed)
+
+    # run filter from phylup
+    before = len(new_seqs)
+
+    f = phyl_up.FilterUniqueAcc(test.config, test.table)
+    f.filter(new_seqs)
+    new_seqs = f.upd_new_seqs
+
+    assert new_seqs.accession.isin(select.accession).all()
+