Spaces:
No application file
No application file
File size: 43,868 Bytes
b7731cd |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 |
# Copyright 2000 by Jeffrey Chang, Brad Chapman. All rights reserved.
# Copyright 2006-2017 by Peter Cock. All rights reserved.
#
# This code is part of the Biopython distribution and governed by its
# license. Please see the LICENSE file that should have been included
# as part of this package.
"""Code to work with GenBank formatted files.
Rather than using Bio.GenBank, you are now encouraged to use Bio.SeqIO with
the "genbank" or "embl" format names to parse GenBank or EMBL files into
SeqRecord and SeqFeature objects (see the Biopython tutorial for details).
Using Bio.GenBank directly to parse GenBank files is only useful if you want
to obtain GenBank-specific Record objects, which is a much closer
representation to the raw file contents than the SeqRecord alternative from
the FeatureParser (used in Bio.SeqIO).
To use the Bio.GenBank parser, there are two helper functions:
- read Parse a handle containing a single GenBank record
as Bio.GenBank specific Record objects.
- parse Iterate over a handle containing multiple GenBank
records as Bio.GenBank specific Record objects.
The following internal classes are not intended for direct use and may
be deprecated in a future release.
Classes:
- Iterator Iterate through a file of GenBank entries
- FeatureParser Parse GenBank data in SeqRecord and SeqFeature objects.
- RecordParser Parse GenBank data into a Record object.
Exceptions:
- ParserFailureError Exception indicating a failure in the parser (ie.
scanner or consumer)
"""
import re
import warnings
from Bio import BiopythonParserWarning
from Bio.Seq import Seq
from Bio.SeqFeature import Location
from Bio.SeqFeature import Reference
from Bio.SeqFeature import SeqFeature
from Bio.SeqFeature import SimpleLocation
from Bio.SeqFeature import LocationParserError
# other Bio.GenBank stuff
from .utils import FeatureValueCleaner
from .Scanner import GenBankScanner
# Constants used to parse GenBank header lines
GENBANK_INDENT = 12
GENBANK_SPACER = " " * GENBANK_INDENT
# Constants for parsing GenBank feature lines
FEATURE_KEY_INDENT = 5
FEATURE_QUALIFIER_INDENT = 21
FEATURE_KEY_SPACER = " " * FEATURE_KEY_INDENT
FEATURE_QUALIFIER_SPACER = " " * FEATURE_QUALIFIER_INDENT
class Iterator:
"""Iterator interface to move over a file of GenBank entries one at a time (OBSOLETE).
This class is likely to be deprecated in a future release of Biopython.
Please use Bio.SeqIO.parse(..., format="gb") or Bio.GenBank.parse(...)
for SeqRecord and GenBank specific Record objects respectively instead.
"""
def __init__(self, handle, parser=None):
"""Initialize the iterator.
Arguments:
- handle - A handle with GenBank entries to iterate through.
- parser - An optional parser to pass the entries through before
returning them. If None, then the raw entry will be returned.
"""
self.handle = handle
self._parser = parser
def __next__(self):
"""Return the next GenBank record from the handle.
Will return None if we ran out of records.
"""
if self._parser is None:
lines = []
while True:
line = self.handle.readline()
if not line:
return None # Premature end of file?
lines.append(line)
if line.rstrip() == "//":
break
return "".join(lines)
try:
return self._parser.parse(self.handle)
except StopIteration:
return None
def __iter__(self):
"""Iterate over the records."""
return iter(self.__next__, None)
class ParserFailureError(ValueError):
"""Failure caused by some kind of problem in the parser."""
pass
_cleaner = FeatureValueCleaner()
class FeatureParser:
"""Parse GenBank files into Seq + Feature objects (OBSOLETE).
Direct use of this class is discouraged, and may be deprecated in
a future release of Biopython.
Please use Bio.SeqIO.parse(...) or Bio.SeqIO.read(...) instead.
"""
def __init__(self, debug_level=0, use_fuzziness=1, feature_cleaner=None):
"""Initialize a GenBank parser and Feature consumer.
Arguments:
- debug_level - An optional argument that species the amount of
debugging information the parser should spit out. By default we have
no debugging info (the fastest way to do things), but if you want
you can set this as high as two and see exactly where a parse fails.
- use_fuzziness - Specify whether or not to use fuzzy representations.
The default is 1 (use fuzziness).
- feature_cleaner - A class which will be used to clean out the
values of features. This class must implement the function
clean_value. GenBank.utils has a "standard" cleaner class, which
is used by default.
"""
self._scanner = GenBankScanner(debug_level)
self.use_fuzziness = use_fuzziness
if feature_cleaner:
self._cleaner = feature_cleaner
else:
self._cleaner = _cleaner # default
def parse(self, handle):
"""Parse the specified handle."""
_consumer = _FeatureConsumer(self.use_fuzziness, self._cleaner)
self._scanner.feed(handle, _consumer)
return _consumer.data
class RecordParser:
"""Parse GenBank files into Record objects (OBSOLETE).
Direct use of this class is discouraged, and may be deprecated in
a future release of Biopython.
Please use the Bio.GenBank.parse(...) or Bio.GenBank.read(...) functions
instead.
"""
def __init__(self, debug_level=0):
"""Initialize the parser.
Arguments:
- debug_level - An optional argument that species the amount of
debugging information the parser should spit out. By default we have
no debugging info (the fastest way to do things), but if you want
you can set this as high as two and see exactly where a parse fails.
"""
self._scanner = GenBankScanner(debug_level)
def parse(self, handle):
"""Parse the specified handle into a GenBank record."""
_consumer = _RecordConsumer()
self._scanner.feed(handle, _consumer)
return _consumer.data
class _BaseGenBankConsumer:
"""Abstract GenBank consumer providing useful general functions (PRIVATE).
This just helps to eliminate some duplication in things that most
GenBank consumers want to do.
"""
# Special keys in GenBank records that we should remove spaces from
# For instance, \translation keys have values which are proteins and
# should have spaces and newlines removed from them. This class
# attribute gives us more control over specific formatting problems.
remove_space_keys = ["translation"]
def __init__(self):
pass
@staticmethod
def _split_keywords(keyword_string):
"""Split a string of keywords into a nice clean list (PRIVATE)."""
# process the keywords into a python list
if keyword_string == "" or keyword_string == ".":
keywords = ""
elif keyword_string[-1] == ".":
keywords = keyword_string[:-1]
else:
keywords = keyword_string
keyword_list = keywords.split(";")
return [x.strip() for x in keyword_list]
@staticmethod
def _split_accessions(accession_string):
"""Split a string of accession numbers into a list (PRIVATE)."""
# first replace all line feeds with spaces
# Also, EMBL style accessions are split with ';'
accession = accession_string.replace("\n", " ").replace(";", " ")
return [x.strip() for x in accession.split() if x.strip()]
@staticmethod
def _split_taxonomy(taxonomy_string):
"""Split a string with taxonomy info into a list (PRIVATE)."""
if not taxonomy_string or taxonomy_string == ".":
# Missing data, no taxonomy
return []
if taxonomy_string[-1] == ".":
tax_info = taxonomy_string[:-1]
else:
tax_info = taxonomy_string
tax_list = tax_info.split(";")
new_tax_list = []
for tax_item in tax_list:
new_items = tax_item.split("\n")
new_tax_list.extend(new_items)
while "" in new_tax_list:
new_tax_list.remove("")
return [x.strip() for x in new_tax_list]
@staticmethod
def _clean_location(location_string):
"""Clean whitespace out of a location string (PRIVATE).
The location parser isn't a fan of whitespace, so we clean it out
before feeding it into the parser.
"""
# Originally this imported string.whitespace and did a replace
# via a loop. It's simpler to just split on whitespace and rejoin
# the string - and this avoids importing string too. See Bug 2684.
return "".join(location_string.split())
@staticmethod
def _remove_newlines(text):
"""Remove any newlines in the passed text, returning the new string (PRIVATE)."""
# get rid of newlines in the qualifier value
newlines = ["\n", "\r"]
for ws in newlines:
text = text.replace(ws, "")
return text
@staticmethod
def _normalize_spaces(text):
"""Replace multiple spaces in the passed text with single spaces (PRIVATE)."""
# get rid of excessive spaces
return " ".join(x for x in text.split(" ") if x)
@staticmethod
def _remove_spaces(text):
"""Remove all spaces from the passed text (PRIVATE)."""
return text.replace(" ", "")
@staticmethod
def _convert_to_python_numbers(start, end):
"""Convert a start and end range to python notation (PRIVATE).
In GenBank, starts and ends are defined in "biological" coordinates,
where 1 is the first base and [i, j] means to include both i and j.
In python, 0 is the first base and [i, j] means to include i, but
not j.
So, to convert "biological" to python coordinates, we need to
subtract 1 from the start, and leave the end and things should
be converted happily.
"""
new_start = start - 1
new_end = end
return new_start, new_end
class _FeatureConsumer(_BaseGenBankConsumer):
"""Create a SeqRecord object with Features to return (PRIVATE).
Attributes:
- use_fuzziness - specify whether or not to parse with fuzziness in
feature locations.
- feature_cleaner - a class that will be used to provide specialized
cleaning-up of feature values.
"""
def __init__(self, use_fuzziness, feature_cleaner=None):
from Bio.SeqRecord import SeqRecord
_BaseGenBankConsumer.__init__(self)
self.data = SeqRecord(None, id=None)
self.data.id = None
self.data.description = ""
self._use_fuzziness = use_fuzziness
self._feature_cleaner = feature_cleaner
self._seq_type = ""
self._seq_data = []
self._cur_reference = None
self._cur_feature = None
self._expected_size = None
def locus(self, locus_name):
"""Set the locus name is set as the name of the Sequence."""
self.data.name = locus_name
def size(self, content):
"""Record the sequence length."""
self._expected_size = int(content)
def residue_type(self, type):
"""Record the sequence type (SEMI-OBSOLETE).
This reflects the fact that the topology (linear/circular) and
molecule type (e.g. DNA vs RNA) were a single field in early
files. Current GenBank/EMBL files have two fields.
"""
self._seq_type = type.strip()
def topology(self, topology):
"""Validate and record sequence topology.
The topology argument should be "linear" or "circular" (string).
"""
if topology:
if topology not in ["linear", "circular"]:
raise ParserFailureError(
f"Unexpected topology {topology!r} should be linear or circular"
)
self.data.annotations["topology"] = topology
def molecule_type(self, mol_type):
"""Validate and record the molecule type (for round-trip etc)."""
if mol_type:
if "circular" in mol_type or "linear" in mol_type:
raise ParserFailureError(
f"Molecule type {mol_type!r} should not include topology"
)
# Writing out records will fail if we have a lower case DNA
# or RNA string in here, so upper case it.
# This is a bit ugly, but we don't want to upper case e.g.
# the m in mRNA, but thanks to the strip we lost the spaces
# so we need to index from the back
if mol_type[-3:].upper() in ("DNA", "RNA") and not mol_type[-3:].isupper():
warnings.warn(
f"Non-upper case molecule type in LOCUS line: {mol_type}",
BiopythonParserWarning,
)
self.data.annotations["molecule_type"] = mol_type
def data_file_division(self, division):
self.data.annotations["data_file_division"] = division
def date(self, submit_date):
self.data.annotations["date"] = submit_date
def definition(self, definition):
"""Set the definition as the description of the sequence."""
if self.data.description:
# Append to any existing description
# e.g. EMBL files with two DE lines.
self.data.description += " " + definition
else:
self.data.description = definition
def accession(self, acc_num):
"""Set the accession number as the id of the sequence.
If we have multiple accession numbers, the first one passed is
used.
"""
new_acc_nums = self._split_accessions(acc_num)
# Also record them ALL in the annotations
try:
# On the off chance there was more than one accession line:
for acc in new_acc_nums:
# Prevent repeat entries
if acc not in self.data.annotations["accessions"]:
self.data.annotations["accessions"].append(acc)
except KeyError:
self.data.annotations["accessions"] = new_acc_nums
# if we haven't set the id information yet, add the first acc num
if not self.data.id:
if len(new_acc_nums) > 0:
# self.data.id = new_acc_nums[0]
# Use the FIRST accession as the ID, not the first on this line!
self.data.id = self.data.annotations["accessions"][0]
def tls(self, content):
self.data.annotations["tls"] = content.split("-")
def tsa(self, content):
self.data.annotations["tsa"] = content.split("-")
def wgs(self, content):
self.data.annotations["wgs"] = content.split("-")
def add_wgs_scafld(self, content):
self.data.annotations.setdefault("wgs_scafld", []).append(content.split("-"))
def nid(self, content):
self.data.annotations["nid"] = content
def pid(self, content):
self.data.annotations["pid"] = content
def version(self, version_id):
# Want to use the versioned accession as the record.id
# This comes from the VERSION line in GenBank files, or the
# obsolete SV line in EMBL. For the new EMBL files we need
# both the version suffix from the ID line and the accession
# from the AC line.
if version_id.count(".") == 1 and version_id.split(".")[1].isdigit():
self.accession(version_id.split(".")[0])
self.version_suffix(version_id.split(".")[1])
elif version_id:
# For backwards compatibility...
self.data.id = version_id
def project(self, content):
"""Handle the information from the PROJECT line as a list of projects.
e.g.::
PROJECT GenomeProject:28471
or::
PROJECT GenomeProject:13543 GenomeProject:99999
This is stored as dbxrefs in the SeqRecord to be consistent with the
projected switch of this line to DBLINK in future GenBank versions.
Note the NCBI plan to replace "GenomeProject:28471" with the shorter
"Project:28471" as part of this transition.
"""
content = content.replace("GenomeProject:", "Project:")
self.data.dbxrefs.extend(p for p in content.split() if p)
def dblink(self, content):
"""Store DBLINK cross references as dbxrefs in our record object.
This line type is expected to replace the PROJECT line in 2009. e.g.
During transition::
PROJECT GenomeProject:28471
DBLINK Project:28471
Trace Assembly Archive:123456
Once the project line is dropped::
DBLINK Project:28471
Trace Assembly Archive:123456
Note GenomeProject -> Project.
We'll have to see some real examples to be sure, but based on the
above example we can expect one reference per line.
Note that at some point the NCBI have included an extra space, e.g.::
DBLINK Project: 28471
"""
# During the transition period with both PROJECT and DBLINK lines,
# we don't want to add the same cross reference twice.
while ": " in content:
content = content.replace(": ", ":")
if content.strip() not in self.data.dbxrefs:
self.data.dbxrefs.append(content.strip())
def version_suffix(self, version):
"""Set the version to overwrite the id.
Since the version provides the same information as the accession
number, plus some extra info, we set this as the id if we have
a version.
"""
# e.g. GenBank line:
# VERSION U49845.1 GI:1293613
# or the obsolete EMBL line:
# SV U49845.1
# Scanner calls consumer.version("U49845.1")
# which then calls consumer.version_suffix(1)
#
# e.g. EMBL new line:
# ID X56734; SV 1; linear; mRNA; STD; PLN; 1859 BP.
# Scanner calls consumer.version_suffix(1)
assert version.isdigit()
self.data.annotations["sequence_version"] = int(version)
def db_source(self, content):
self.data.annotations["db_source"] = content.rstrip()
def gi(self, content):
self.data.annotations["gi"] = content
def keywords(self, content):
if "keywords" in self.data.annotations:
# Multi-line keywords, append to list
# Note EMBL states "A keyword is never split between lines."
self.data.annotations["keywords"].extend(self._split_keywords(content))
else:
self.data.annotations["keywords"] = self._split_keywords(content)
def segment(self, content):
self.data.annotations["segment"] = content
def source(self, content):
# Note that some software (e.g. VectorNTI) may produce an empty
# source (rather than using a dot/period as might be expected).
if content == "":
source_info = ""
elif content[-1] == ".":
source_info = content[:-1]
else:
source_info = content
self.data.annotations["source"] = source_info
def organism(self, content):
self.data.annotations["organism"] = content
def taxonomy(self, content):
"""Record (another line of) the taxonomy lineage."""
lineage = self._split_taxonomy(content)
try:
self.data.annotations["taxonomy"].extend(lineage)
except KeyError:
self.data.annotations["taxonomy"] = lineage
def reference_num(self, content):
"""Signal the beginning of a new reference object."""
# if we have a current reference that hasn't been added to
# the list of references, add it.
if self._cur_reference is not None:
self.data.annotations["references"].append(self._cur_reference)
else:
self.data.annotations["references"] = []
self._cur_reference = Reference()
def reference_bases(self, content):
"""Attempt to determine the sequence region the reference entails.
Possible types of information we may have to deal with:
(bases 1 to 86436)
(sites)
(bases 1 to 105654; 110423 to 111122)
1 (residues 1 to 182)
"""
# first remove the parentheses
assert content.endswith(")"), content
ref_base_info = content[1:-1]
all_locations = []
# parse if we've got 'bases' and 'to'
if "bases" in ref_base_info and "to" in ref_base_info:
# get rid of the beginning 'bases'
ref_base_info = ref_base_info[5:]
locations = self._split_reference_locations(ref_base_info)
all_locations.extend(locations)
elif "residues" in ref_base_info and "to" in ref_base_info:
residues_start = ref_base_info.find("residues")
# get only the information after "residues"
ref_base_info = ref_base_info[(residues_start + len("residues ")) :]
locations = self._split_reference_locations(ref_base_info)
all_locations.extend(locations)
# make sure if we are not finding information then we have
# the string 'sites' or the string 'bases'
elif ref_base_info == "sites" or ref_base_info.strip() == "bases":
pass
# otherwise raise an error
else:
raise ValueError(
f"Could not parse base info {ref_base_info} in record {self.data.id}"
)
self._cur_reference.location = all_locations
def _split_reference_locations(self, location_string):
"""Get reference locations out of a string of reference information (PRIVATE).
The passed string should be of the form::
1 to 20; 20 to 100
This splits the information out and returns a list of location objects
based on the reference locations.
"""
# split possibly multiple locations using the ';'
all_base_info = location_string.split(";")
new_locations = []
for base_info in all_base_info:
start, end = base_info.split("to")
new_start, new_end = self._convert_to_python_numbers(
int(start.strip()), int(end.strip())
)
this_location = SimpleLocation(new_start, new_end)
new_locations.append(this_location)
return new_locations
def authors(self, content):
if self._cur_reference.authors:
self._cur_reference.authors += " " + content
else:
self._cur_reference.authors = content
def consrtm(self, content):
if self._cur_reference.consrtm:
self._cur_reference.consrtm += " " + content
else:
self._cur_reference.consrtm = content
def title(self, content):
if self._cur_reference is None:
warnings.warn(
"GenBank TITLE line without REFERENCE line.", BiopythonParserWarning
)
elif self._cur_reference.title:
self._cur_reference.title += " " + content
else:
self._cur_reference.title = content
def journal(self, content):
if self._cur_reference.journal:
self._cur_reference.journal += " " + content
else:
self._cur_reference.journal = content
def medline_id(self, content):
self._cur_reference.medline_id = content
def pubmed_id(self, content):
self._cur_reference.pubmed_id = content
def remark(self, content):
"""Deal with a reference comment."""
if self._cur_reference.comment:
self._cur_reference.comment += " " + content
else:
self._cur_reference.comment = content
def comment(self, content):
try:
self.data.annotations["comment"] += "\n" + "\n".join(content)
except KeyError:
self.data.annotations["comment"] = "\n".join(content)
def structured_comment(self, content):
self.data.annotations["structured_comment"] = content
def features_line(self, content):
"""Get ready for the feature table when we reach the FEATURE line."""
self.start_feature_table()
def start_feature_table(self):
"""Indicate we've got to the start of the feature table."""
# make sure we've added on our last reference object
if self._cur_reference is not None:
self.data.annotations["references"].append(self._cur_reference)
self._cur_reference = None
def feature_key(self, content):
# start a new feature
self._cur_feature = SeqFeature()
self._cur_feature.type = content
self.data.features.append(self._cur_feature)
def location(self, content):
"""Parse out location information from the location string.
This uses simple Python code with some regular expressions to do the
parsing, and then translates the results into appropriate objects.
"""
# clean up newlines and other whitespace inside the location before
# parsing - locations should have no whitespace whatsoever
location_line = self._clean_location(content)
# Older records have junk like replace(266,"c") in the
# location line. Newer records just replace this with
# the number 266 and have the information in a more reasonable
# place. So we'll just grab out the number and feed this to the
# parser. We shouldn't really be losing any info this way.
if "replace" in location_line:
comma_pos = location_line.find(",")
location_line = location_line[8:comma_pos]
length = self._expected_size
# Check if the sequence is circular for features that span the origin
is_circular = "circular" in self.data.annotations.get("topology", "").lower()
stranded = "PROTEIN" not in self._seq_type.upper()
try:
location = Location.fromstring(location_line, length, is_circular, stranded)
except LocationParserError as e:
warnings.warn(
f"{e}; setting feature location to None.", BiopythonParserWarning
)
location = None
self._cur_feature.location = location
def feature_qualifier(self, key, value):
"""When we get a qualifier key and its value.
Can receive None, since you can have valueless keys such as /pseudo
"""
# Hack to try to preserve historical behaviour of /pseudo etc
if value is None:
# if the key doesn't exist yet, add an empty string
if key not in self._cur_feature.qualifiers:
self._cur_feature.qualifiers[key] = [""]
return
# otherwise just skip this key
return
# Remove enclosing quotation marks
if len(value) > 1 and value[0] == '"' and value[-1] == '"':
value = value[1:-1]
# Handle NCBI escaping
# Warn if escaping is not according to standard
if re.search(r'[^"]"[^"]|^"[^"]|[^"]"$', value):
warnings.warn(
'The NCBI states double-quote characters like " should be escaped as "" '
"(two double - quotes), but here it was not: %r" % value,
BiopythonParserWarning,
)
# Undo escaping, repeated double quotes -> one double quote
value = value.replace('""', '"')
if self._feature_cleaner is not None:
value = self._feature_cleaner.clean_value(key, value)
# if the qualifier name exists, append the value
if key in self._cur_feature.qualifiers:
self._cur_feature.qualifiers[key].append(value)
# otherwise start a new list of the key with its values
else:
self._cur_feature.qualifiers[key] = [value]
def feature_qualifier_name(self, content_list):
"""Use feature_qualifier instead (OBSOLETE)."""
raise NotImplementedError("Use the feature_qualifier method instead.")
def feature_qualifier_description(self, content):
"""Use feature_qualifier instead (OBSOLETE)."""
raise NotImplementedError("Use the feature_qualifier method instead.")
def contig_location(self, content):
"""Deal with CONTIG information."""
# Historically this was stored as a SeqFeature object, but it was
# stored under record.annotations["contig"] and not under
# record.features with the other SeqFeature objects.
#
# The CONTIG location line can include additional tokens like
# Gap(), Gap(100) or Gap(unk100) which are not used in the feature
# location lines, so storing it using SeqFeature based location
# objects is difficult.
#
# We now store this a string, which means for BioSQL we are now in
# much better agreement with how BioPerl records the CONTIG line
# in the database.
#
# NOTE - This code assumes the scanner will return all the CONTIG
# lines already combined into one long string!
self.data.annotations["contig"] = content
def origin_name(self, content):
pass
def base_count(self, content):
pass
def base_number(self, content):
pass
def sequence(self, content):
"""Add up sequence information as we get it.
To try and make things speedier, this puts all of the strings
into a list of strings, and then uses string.join later to put
them together. Supposedly, this is a big time savings
"""
assert " " not in content
self._seq_data.append(content.upper())
def record_end(self, content):
"""Clean up when we've finished the record."""
# Try and append the version number to the accession for the full id
if not self.data.id:
if "accessions" in self.data.annotations:
raise ValueError(
"Problem adding version number to accession: "
+ str(self.data.annotations["accessions"])
)
self.data.id = self.data.name # Good fall back?
elif self.data.id.count(".") == 0:
try:
self.data.id += ".%i" % self.data.annotations["sequence_version"]
except KeyError:
pass
# add the sequence information
sequence = "".join(self._seq_data)
if (
self._expected_size is not None
and len(sequence) != 0
and self._expected_size != len(sequence)
):
warnings.warn(
"Expected sequence length %i, found %i (%s)."
% (self._expected_size, len(sequence), self.data.id),
BiopythonParserWarning,
)
molecule_type = None
if self._seq_type:
# mRNA is really also DNA, since it is actually cDNA
if "DNA" in self._seq_type.upper() or "MRNA" in self._seq_type.upper():
molecule_type = "DNA"
# are there ever really RNA sequences in GenBank?
elif "RNA" in self._seq_type.upper():
# Even for data which was from RNA, the sequence string
# is usually given as DNA (T not U). Bug 3010
molecule_type = "RNA"
elif (
"PROTEIN" in self._seq_type.upper() or self._seq_type == "PRT"
): # PRT is used in EMBL-bank for patents
molecule_type = "protein"
# work around ugly GenBank records which have circular or
# linear but no indication of sequence type
elif self._seq_type in ["circular", "linear", "unspecified"]:
pass
# we have a bug if we get here
else:
raise ValueError(
f"Could not determine molecule_type for seq_type {self._seq_type}"
)
# Don't overwrite molecule_type
if molecule_type is not None:
self.data.annotations["molecule_type"] = self.data.annotations.get(
"molecule_type", molecule_type
)
if not sequence and self._expected_size:
self.data.seq = Seq(None, length=self._expected_size)
else:
self.data.seq = Seq(sequence)
class _RecordConsumer(_BaseGenBankConsumer):
"""Create a GenBank Record object from scanner generated information (PRIVATE)."""
def __init__(self):
_BaseGenBankConsumer.__init__(self)
from . import Record
self.data = Record.Record()
self._seq_data = []
self._cur_reference = None
self._cur_feature = None
self._cur_qualifier = None
def tls(self, content):
self.data.tls = content.split("-")
def tsa(self, content):
self.data.tsa = content.split("-")
def wgs(self, content):
self.data.wgs = content.split("-")
def add_wgs_scafld(self, content):
self.data.wgs_scafld.append(content.split("-"))
def locus(self, content):
self.data.locus = content
def size(self, content):
self.data.size = content
def residue_type(self, content):
# Be lenient about parsing, but technically lowercase residue types are malformed.
if "dna" in content or "rna" in content:
warnings.warn(
f"Invalid seq_type ({content}): DNA/RNA should be uppercase.",
BiopythonParserWarning,
)
self.data.residue_type = content
def data_file_division(self, content):
self.data.data_file_division = content
def date(self, content):
self.data.date = content
def definition(self, content):
self.data.definition = content
def accession(self, content):
for acc in self._split_accessions(content):
if acc not in self.data.accession:
self.data.accession.append(acc)
def molecule_type(self, mol_type):
"""Validate and record the molecule type (for round-trip etc)."""
if mol_type:
if "circular" in mol_type or "linear" in mol_type:
raise ParserFailureError(
f"Molecule type {mol_type!r} should not include topology"
)
# Writing out records will fail if we have a lower case DNA
# or RNA string in here, so upper case it.
# This is a bit ugly, but we don't want to upper case e.g.
# the m in mRNA, but thanks to the strip we lost the spaces
# so we need to index from the back
if mol_type[-3:].upper() in ("DNA", "RNA") and not mol_type[-3:].isupper():
warnings.warn(
f"Non-upper case molecule type in LOCUS line: {mol_type}",
BiopythonParserWarning,
)
self.data.molecule_type = mol_type
def topology(self, topology):
"""Validate and record sequence topology.
The topology argument should be "linear" or "circular" (string).
"""
if topology:
if topology not in ["linear", "circular"]:
raise ParserFailureError(
f"Unexpected topology {topology!r} should be linear or circular"
)
self.data.topology = topology
def nid(self, content):
self.data.nid = content
def pid(self, content):
self.data.pid = content
def version(self, content):
self.data.version = content
def db_source(self, content):
self.data.db_source = content.rstrip()
def gi(self, content):
self.data.gi = content
def keywords(self, content):
self.data.keywords = self._split_keywords(content)
def project(self, content):
self.data.projects.extend(p for p in content.split() if p)
def dblink(self, content):
self.data.dblinks.append(content)
def segment(self, content):
self.data.segment = content
def source(self, content):
self.data.source = content
def organism(self, content):
self.data.organism = content
def taxonomy(self, content):
self.data.taxonomy = self._split_taxonomy(content)
def reference_num(self, content):
"""Grab the reference number and signal the start of a new reference."""
# check if we have a reference to add
if self._cur_reference is not None:
self.data.references.append(self._cur_reference)
from . import Record
self._cur_reference = Record.Reference()
self._cur_reference.number = content
def reference_bases(self, content):
self._cur_reference.bases = content
def authors(self, content):
self._cur_reference.authors = content
def consrtm(self, content):
self._cur_reference.consrtm = content
def title(self, content):
if self._cur_reference is None:
warnings.warn(
"GenBank TITLE line without REFERENCE line.", BiopythonParserWarning
)
return
self._cur_reference.title = content
def journal(self, content):
self._cur_reference.journal = content
def medline_id(self, content):
self._cur_reference.medline_id = content
def pubmed_id(self, content):
self._cur_reference.pubmed_id = content
def remark(self, content):
self._cur_reference.remark = content
def comment(self, content):
self.data.comment += "\n".join(content)
def structured_comment(self, content):
self.data.structured_comment = content
def primary_ref_line(self, content):
"""Save reference data for the PRIMARY line."""
self.data.primary.append(content)
def primary(self, content):
pass
def features_line(self, content):
"""Get ready for the feature table when we reach the FEATURE line."""
self.start_feature_table()
def start_feature_table(self):
"""Signal the start of the feature table."""
# we need to add on the last reference
if self._cur_reference is not None:
self.data.references.append(self._cur_reference)
def feature_key(self, content):
"""Grab the key of the feature and signal the start of a new feature."""
# first add on feature information if we've got any
self._add_feature()
from . import Record
self._cur_feature = Record.Feature()
self._cur_feature.key = content
def _add_feature(self):
"""Add a feature to the record, with relevant checks (PRIVATE).
This does all of the appropriate checking to make sure we haven't
left any info behind, and that we are only adding info if it
exists.
"""
if self._cur_feature is not None:
# if we have a left over qualifier, add it to the qualifiers
# on the current feature
if self._cur_qualifier is not None:
self._cur_feature.qualifiers.append(self._cur_qualifier)
self._cur_qualifier = None
self.data.features.append(self._cur_feature)
def location(self, content):
self._cur_feature.location = self._clean_location(content)
def feature_qualifier(self, key, value):
self.feature_qualifier_name([key])
if value is not None:
self.feature_qualifier_description(value)
def feature_qualifier_name(self, content_list):
"""Deal with qualifier names.
We receive a list of keys, since you can have valueless keys such as
/pseudo which would be passed in with the next key (since no other
tags separate them in the file)
"""
from . import Record
for content in content_list:
# the record parser keeps the /s -- add them if we don't have 'em
if not content.startswith("/"):
content = f"/{content}"
# add on a qualifier if we've got one
if self._cur_qualifier is not None:
self._cur_feature.qualifiers.append(self._cur_qualifier)
self._cur_qualifier = Record.Qualifier()
self._cur_qualifier.key = content
def feature_qualifier_description(self, content):
# if we have info then the qualifier key should have a ='s
if "=" not in self._cur_qualifier.key:
self._cur_qualifier.key = f"{self._cur_qualifier.key}="
cur_content = self._remove_newlines(content)
# remove all spaces from the value if it is a type where spaces
# are not important
for remove_space_key in self.__class__.remove_space_keys:
if remove_space_key in self._cur_qualifier.key:
cur_content = self._remove_spaces(cur_content)
self._cur_qualifier.value = self._normalize_spaces(cur_content)
def base_count(self, content):
self.data.base_counts = content
def origin_name(self, content):
self.data.origin = content
def contig_location(self, content):
"""Signal that we have contig information to add to the record."""
self.data.contig = self._clean_location(content)
def sequence(self, content):
"""Add sequence information to a list of sequence strings.
This removes spaces in the data and uppercases the sequence, and
then adds it to a list of sequences. Later on we'll join this
list together to make the final sequence. This is faster than
adding on the new string every time.
"""
assert " " not in content
self._seq_data.append(content.upper())
def record_end(self, content):
"""Signal the end of the record and do any necessary clean-up."""
# add together all of the sequence parts to create the
# final sequence string
self.data.sequence = "".join(self._seq_data)
# add on the last feature
self._add_feature()
def parse(handle):
"""Iterate over GenBank formatted entries as Record objects.
>>> from Bio import GenBank
>>> with open("GenBank/NC_000932.gb") as handle:
... for record in GenBank.parse(handle):
... print(record.accession)
['NC_000932']
To get SeqRecord objects use Bio.SeqIO.parse(..., format="gb")
instead.
"""
return iter(Iterator(handle, RecordParser()))
def read(handle):
"""Read a handle containing a single GenBank entry as a Record object.
>>> from Bio import GenBank
>>> with open("GenBank/NC_000932.gb") as handle:
... record = GenBank.read(handle)
... print(record.accession)
['NC_000932']
To get a SeqRecord object use Bio.SeqIO.read(..., format="gb")
instead.
"""
iterator = parse(handle)
try:
record = next(iterator)
except StopIteration:
raise ValueError("No records found in handle") from None
try:
next(iterator)
raise ValueError("More than one record found in handle")
except StopIteration:
pass
return record
if __name__ == "__main__":
from Bio._utils import run_doctest
run_doctest()
|