jbrowse: gff3_rebase.py annotate

annotate gff3_rebase.py @ 33:0ae74c70b267 draft

"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 369a727966d697c56633b27ad2757db40fef0dc3"

author	iuc
date	Fri, 30 Aug 2019 03:36:39 -0400
parents	ff11d442feed
children

rev	line source
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	1 #!/usr/bin/env python
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	2 import argparse
3 7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	3 import copy
8 ad4b9d7eae6a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 9a243c616a4a3156347e38fdb5f35863ae5133f9 iuc parents: 3 diff changeset	4 import logging
ad4b9d7eae6a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 9a243c616a4a3156347e38fdb5f35863ae5133f9 iuc parents: 3 diff changeset	5 import sys
ad4b9d7eae6a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 9a243c616a4a3156347e38fdb5f35863ae5133f9 iuc parents: 3 diff changeset	6
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	7 from BCBio import GFF
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	8 from Bio.SeqFeature import FeatureLocation
8 ad4b9d7eae6a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 9a243c616a4a3156347e38fdb5f35863ae5133f9 iuc parents: 3 diff changeset	9
ad4b9d7eae6a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 9a243c616a4a3156347e38fdb5f35863ae5133f9 iuc parents: 3 diff changeset	10 logging.basicConfig(level=logging.INFO)
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	11 log = logging.getLogger(__name__)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	12
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	13 __author__ = "Eric Rasche"
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	14 __version__ = "0.4.0"
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	15 __maintainer__ = "Eric Rasche"
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	16 __email__ = "esr@tamu.edu"
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	17
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	18
3 7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	19 def feature_lambda(feature_list, test, test_kwargs, subfeatures=True):
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	20 """Recursively search through features, testing each with a test function, yielding matches.
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	21
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	22 GFF3 is a hierachical data structure, so we need to be able to recursively
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	23 search through features. E.g. if you're looking for a feature with
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	24 ID='bob.42', you can't just do a simple list comprehension with a test
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	25 case. You don't know how deeply burried bob.42 will be in the feature tree. This is where feature_lambda steps in.
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	26
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	27 :type feature_list: list
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	28 :param feature_list: an iterable of features
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	29
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	30 :type test: function reference
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	31 :param test: a closure with the method signature (feature, **kwargs) where
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	32 the kwargs are those passed in the next argument. This
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	33 function should return True or False, True if the feature is
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	34 to be yielded as part of the main feature_lambda function, or
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	35 False if it is to be ignored. This function CAN mutate the
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	36 features passed to it (think "apply").
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	37
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	38 :type test_kwargs: dictionary
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	39 :param test_kwargs: kwargs to pass to your closure when it is called.
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	40
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	41 :type subfeatures: boolean
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	42 :param subfeatures: when a feature is matched, should just that feature be
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	43 yielded to the caller, or should the entire sub_feature
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	44 tree for that feature be included? subfeatures=True is
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	45 useful in cases such as searching for a gene feature,
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	46 and wanting to know what RBS/Shine_Dalgarno_sequences
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	47 are in the sub_feature tree (which can be accomplished
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	48 with two feature_lambda calls). subfeatures=False is
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	49 useful in cases when you want to process (and possibly
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	50 return) the entire feature tree, such as applying a
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	51 qualifier to every single feature.
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	52
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	53 :rtype: yielded list
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	54 :return: Yields a list of matching features.
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	55 """
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	56 # Either the top level set of [features] or the subfeature attribute
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	57 for feature in feature_list:
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	58 if test(feature, **test_kwargs):
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	59 if not subfeatures:
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	60 feature_copy = copy.deepcopy(feature)
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	61 feature_copy.sub_features = []
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	62 yield feature_copy
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	63 else:
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	64 yield feature
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	65
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	66 if hasattr(feature, 'sub_features'):
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	67 for x in feature_lambda(feature.sub_features, test, test_kwargs, subfeatures=subfeatures):
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	68 yield x
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	69
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	70
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	71 def feature_test_qual_value(feature, **kwargs):
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	72 """Test qualifier values.
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	73
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	74 For every feature, check that at least one value in
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	75 feature.quailfiers(kwargs['qualifier']) is in kwargs['attribute_list']
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	76 """
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	77 for attribute_value in feature.qualifiers.get(kwargs['qualifier'], []):
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	78 if attribute_value in kwargs['attribute_list']:
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	79 return True
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	80 return False
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	81
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	82
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	83 def __get_features(child, interpro=False):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	84 child_features = {}
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	85 for rec in GFF.parse(child):
17 ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	86 # Only top level
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	87 for feature in rec.features:
17 ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	88 # Get the record id as parent_feature_id (since this is how it will be during remapping)
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	89 parent_feature_id = rec.id
17 ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	90 # If it's an interpro specific gff3 file
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	91 if interpro:
17 ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	92 # Then we ignore polypeptide features as they're useless
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	93 if feature.type == 'polypeptide':
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	94 continue
17 ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	95 # If there's an underscore, we strip up to that underscore?
ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	96 # I do not know the rationale for this, removing.
ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	97 # if '_' in parent_feature_id:
ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	98 # parent_feature_id = parent_feature_id[parent_feature_id.index('_') + 1:]
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	99
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	100 try:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	101 child_features[parent_feature_id].append(feature)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	102 except KeyError:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	103 child_features[parent_feature_id] = [feature]
17 ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	104 # Keep a list of feature objects keyed by parent record id
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	105 return child_features
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	106
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	107
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	108 def __update_feature_location(feature, parent, protein2dna):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	109 start = feature.location.start
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	110 end = feature.location.end
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	111 if protein2dna:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	112 start *= 3
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	113 end *= 3
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	114
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	115 if parent.location.strand >= 0:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	116 ns = parent.location.start + start
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	117 ne = parent.location.start + end
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	118 st = +1
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	119 else:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	120 ns = parent.location.end - end
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	121 ne = parent.location.end - start
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	122 st = -1
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	123
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	124 # Don't let start/stops be less than zero. It's technically valid for them
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	125 # to be (at least in the model I'm working with) but it causes numerous
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	126 # issues.
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	127 #
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	128 # Instead, we'll replace with %3 to try and keep it in the same reading
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	129 # frame that it should be in.
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	130 if ns < 0:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	131 ns %= 3
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	132 if ne < 0:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	133 ne %= 3
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	134
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	135 feature.location = FeatureLocation(ns, ne, strand=st)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	136
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	137 if hasattr(feature, 'sub_features'):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	138 for subfeature in feature.sub_features:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	139 __update_feature_location(subfeature, parent, protein2dna)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	140
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	141
17 ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	142 def rebase(parent, child, interpro=False, protein2dna=False, map_by='ID'):
ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	143 # get all of the features we will be re-mapping in a dictionary, keyed by parent feature ID
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	144 child_features = __get_features(child, interpro=interpro)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	145
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	146 for rec in GFF.parse(parent):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	147 replacement_features = []
3 7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	148 for feature in feature_lambda(
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	149 rec.features,
17 ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	150 # Filter features in the parent genome by those that are
ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	151 # "interesting", i.e. have results in child_features array.
ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	152 # Probably an unnecessary optimisation.
3 7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	153 feature_test_qual_value,
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	154 {
17 ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	155 'qualifier': map_by,
3 7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	156 'attribute_list': child_features.keys(),
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	157 },
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	158 subfeatures=False):
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	159
17 ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	160 # Features which will be re-mapped
ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	161 to_remap = child_features[feature.id]
ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	162 # TODO: update starts
ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	163 fixed_features = []
ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	164 for x in to_remap:
3 7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	165 # Then update the location of the actual feature
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	166 __update_feature_location(x, feature, protein2dna)
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	167
3 7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	168 if interpro:
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	169 for y in ('status', 'Target'):
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	170 try:
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	171 del x.qualifiers[y]
17 ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	172 except Exception:
3 7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	173 pass
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	174
17 ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	175 fixed_features.append(x)
ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	176 replacement_features.extend(fixed_features)
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	177 # We do this so we don't include the original set of features that we
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	178 # were rebasing against in our result.
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	179 rec.features = replacement_features
3 7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	180 rec.annotations = {}
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	181 GFF.write([rec], sys.stdout)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	182
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	183
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	184 if __name__ == '__main__':
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	185 parser = argparse.ArgumentParser(description='rebase gff3 features against parent locations', epilog="")
14 18be2d72fdee planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0bfb0de98c918860c21808e8832caad9f0535975 iuc parents: 8 diff changeset	186 parser.add_argument('parent', type=argparse.FileType('r'), help='Parent GFF3 annotations')
18be2d72fdee planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0bfb0de98c918860c21808e8832caad9f0535975 iuc parents: 8 diff changeset	187 parser.add_argument('child', type=argparse.FileType('r'), help='Child GFF3 annotations to rebase against parent')
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	188 parser.add_argument('--interpro', action='store_true',
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	189 help='Interpro specific modifications')
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	190 parser.add_argument('--protein2dna', action='store_true',
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	191 help='Map protein translated results to original DNA data')
17 ff11d442feed planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 908f16ea4eb082227437dc93e06e8cb742f5a257 iuc parents: 14 diff changeset	192 parser.add_argument('--map_by', help='Map by key', default='ID')
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	193 args = parser.parse_args()
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	194 rebase(**vars(args))

Mercurial > repos > iuc > jbrowse

annotate gff3_rebase.py @ 33:0ae74c70b267 draft