jbrowse: gff3_rebase.py annotate

annotate gff3_rebase.py @ 5:ae9382cfb6ac draft

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 3bbca939ca8a3298a3a2d6450abb04a96851e1ed

author	iuc
date	Sat, 25 Jun 2016 15:06:43 -0400
parents	7342f467507b
children	ad4b9d7eae6a

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 sys
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	3 import logging
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	4 logging.basicConfig(level=logging.INFO)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	5 import argparse
3 7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	6 import copy
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
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	9 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	10
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	11 __author__ = "Eric Rasche"
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	12 __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	13 __maintainer__ = "Eric Rasche"
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	14 __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	15
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	16
3 7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	17 def feature_lambda(feature_list, test, test_kwargs, subfeatures=True):
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	18 """Recursively search through features, testing each with a test function, yielding matches.
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	19
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	20 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	21 search through features. E.g. if you're looking for a feature with
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	22 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	23 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	24
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	25 :type feature_list: list
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	26 :param feature_list: an iterable of features
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	27
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	28 :type test: function reference
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	29 :param test: a closure with the method signature (feature, **kwargs) where
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	30 the kwargs are those passed in the next argument. This
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	31 function should return True or False, True if the feature is
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	32 to be yielded as part of the main feature_lambda function, or
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	33 False if it is to be ignored. This function CAN mutate the
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	34 features passed to it (think "apply").
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	35
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	36 :type test_kwargs: dictionary
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	37 :param test_kwargs: kwargs to pass to your closure when it is called.
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	38
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	39 :type subfeatures: boolean
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	40 :param subfeatures: when a feature is matched, should just that feature be
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	41 yielded to the caller, or should the entire sub_feature
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	42 tree for that feature be included? subfeatures=True is
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	43 useful in cases such as searching for a gene feature,
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	44 and wanting to know what RBS/Shine_Dalgarno_sequences
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	45 are in the sub_feature tree (which can be accomplished
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	46 with two feature_lambda calls). subfeatures=False is
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	47 useful in cases when you want to process (and possibly
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	48 return) the entire feature tree, such as applying a
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	49 qualifier to every single feature.
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	50
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	51 :rtype: yielded list
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	52 :return: Yields a list of matching features.
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	53 """
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	54 # Either the top level set of [features] or the subfeature attribute
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	55 for feature in feature_list:
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	56 if test(feature, **test_kwargs):
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	57 if not subfeatures:
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	58 feature_copy = copy.deepcopy(feature)
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	59 feature_copy.sub_features = []
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	60 yield feature_copy
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	61 else:
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	62 yield feature
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	63
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	64 if hasattr(feature, 'sub_features'):
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	65 for x in feature_lambda(feature.sub_features, test, test_kwargs, subfeatures=subfeatures):
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	66 yield x
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	67
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	68
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	69 def feature_test_qual_value(feature, **kwargs):
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	70 """Test qualifier values.
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	71
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	72 For every feature, check that at least one value in
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	73 feature.quailfiers(kwargs['qualifier']) is in kwargs['attribute_list']
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	74 """
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	75 for attribute_value in feature.qualifiers.get(kwargs['qualifier'], []):
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	76 if attribute_value in kwargs['attribute_list']:
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	77 return True
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	78 return False
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	79
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	80
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	81 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	82 child_features = {}
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	83 for rec in GFF.parse(child):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	84 for feature in rec.features:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	85 parent_feature_id = rec.id
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	86 if interpro:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	87 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	88 continue
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	89 if '_' in parent_feature_id:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	90 parent_feature_id = parent_feature_id[parent_feature_id.index('_') + 1:]
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	91
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	92 try:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	93 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	94 except KeyError:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	95 child_features[parent_feature_id] = [feature]
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	96 return child_features
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	97
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	98
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	99 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	100 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	101 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	102 if protein2dna:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	103 start *= 3
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	104 end *= 3
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	105
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	106 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	107 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	108 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	109 st = +1
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	110 else:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	111 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	112 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	113 st = -1
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 # 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	116 # 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	117 # issues.
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	118 #
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	119 # 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	120 # 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	121 if ns < 0:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	122 ns %= 3
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	123 if ne < 0:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	124 ne %= 3
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	125
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	126 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	127
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	128 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	129 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	130 __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	131
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	132
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	133 def rebase(parent, child, interpro=False, protein2dna=False):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	134 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	135
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	136 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	137 replacement_features = []
3 7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	138 for feature in feature_lambda(
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	139 rec.features,
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	140 feature_test_qual_value,
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	141 {
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	142 'qualifier': 'ID',
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	143 'attribute_list': child_features.keys(),
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	144 },
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	145 subfeatures=False):
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	146
3 7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	147 new_subfeatures = child_features[feature.id]
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	148 fixed_subfeatures = []
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	149 for x in new_subfeatures:
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	150 # Then update the location of the actual feature
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	151 __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	152
3 7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	153 if interpro:
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	154 for y in ('status', 'Target'):
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	155 try:
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	156 del x.qualifiers[y]
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	157 except:
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	158 pass
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	159
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	160 fixed_subfeatures.append(x)
7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	161 replacement_features.extend(fixed_subfeatures)
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	162 # 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	163 # 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	164 rec.features = replacement_features
3 7342f467507b Uploaded v0.4 of JBrowse iuc parents: 1 diff changeset	165 rec.annotations = {}
1 497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	166 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	167
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	168
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	169 if __name__ == '__main__':
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	170 parser = argparse.ArgumentParser(description='rebase gff3 features against parent locations', epilog="")
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	171 parser.add_argument('parent', type=file, help='Parent GFF3 annotations')
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	172 parser.add_argument('child', help='Child GFF3 annotations to rebase against parent')
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	173 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	174 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	175 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	176 help='Map protein translated results to original DNA data')
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f iuc parents: diff changeset	177 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	178 rebase(**vars(args))

Mercurial > repos > iuc > jbrowse

annotate gff3_rebase.py @ 5:ae9382cfb6ac draft