def make_2id(s, dbname, primary_name, secondary_name):
assert secondary_name is not None
if primary_name is None:
return Str(s + "||") + \
Std.dbxref_dbid(UntilSep(sep = "| "), {"dbname": dbname,
"type": secondary_name})
return Str(s + "|") + \
Std.dbxref_dbid(UntilSep(sep = "|"), {"dbname": dbname,
"type": primary_name}) + \
Str("|") + \
Std.dbxref_dbid(UntilSep(sep = "| "), {"dbname": dbname,
"type": secondary_name})
Martel.AnyEol() + \
Martel.Str("FH") + \
Martel.AnyEol()
## FT - feature table data (>=0 per entry)
##FT_line = Martel.Str("FT ") + \
## Martel.ToEol("ft_data")
##FT_block = Martel.Rep1(FT_line)
fq_dbxref = Std.feature_qualifier_name(Martel.Str("db_xref")) + \
Martel.Str('=') + \
Std.feature_qualifier_description(
Martel.Str('"') + \
Std.dbxref(Std.dbxref_dbname(Martel.UntilSep(None, ":")) + \
Martel.Str(":") + \
Std.dbxref_dbid(Martel.UntilSep(None, '"'))) + \
Martel.Str('"')) + \
Martel.AnyEol()
fq_generic = \
Martel.Assert(Martel.Word() + Martel.Str("=")) + \
Std.feature_qualifier_name(Martel.Word()) + \
Martel.Str("=") + \
Std.feature_qualifier_description(Martel.UntilEol()) + \
Martel.AnyEol() + \
Martel.Rep(
Martel.Str("FT ") + \
(Martel.AssertNot(Martel.Str("/")) |
Martel.AssertNot(Martel.Re(r"/\w+="))) + \
Std.feature_qualifier_description(Martel.UntilEol()) + \
)
)
#--- DR
# This is needed for things like
# DR MGD; MGI:95401; EPB4.1.
# where I need to scan up to the last "." That is, I want
# "EPB4.1" to be the secondary identifier, not "EPB4" nor "EPB4.1."
_to_secondary_end = Martel.Re(r"([^.\R]|(?!.\R)\.)+")
database_id = Std.dbxref_dbname(Martel.UntilSep("database_identifier", ";"),
{"style": "sp"})
primary_id = Std.dbxref_dbid(Martel.UntilSep("primary_identifier", ";"),
{"type": "primary"})
secondary_id = Std.dbxref_dbid(Martel.Group("secondary_identifier",
_to_secondary_end),
{"type": "accession"})
# used in StdHandler for fast dxbref - don't rename!
real_DR_general = Std.dbxref(database_id + Martel.Str("; ") + \
primary_id + Martel.Str("; ") + \
secondary_id,
)
fast_DR_general = Std.fast_dbxref(real_DR_general,
{"style": "sp-general"})
DR_general = Martel.FastFeature(fast_DR_general, "fast-sp-dbxref",
real_DR_general.group_names() )
def make_1id(s, dbname, name):
return Str(s + "|") + \
Std.dbxref_dbid(UntilSep(sep = "| "), {"dbname": dbname,
"type": name})
ids.append(make_2id("sp", "sp", "primary", "secondary"))
# Brookhaven Protein Data Bank pdb|entry|chain
ids.append(make_2id("pdb", "x-pdb", "primary", "secondary")) # XXX not correct
# Patents pat|country|number
ids.append(make_2id("pat", "x-pat", "primary", "secondary")) # XXX not correct
# GenInfo Backbone Id bbs|number
ids.append(make_1id("bbs", "x-bbs", "primary"))
# General database identifier gnl|database|identifier
gnl_id = Str("gnl|") + \
Std.dbxref_dbname(UntilSep(sep = "| ")) + \
Str("|") + \
Std.dbxref_dbid(UntilSep(sep = "| "))
ids.append(gnl_id)
# NCBI Reference Sequence ref|accession|locus
ids.append(make_2id("ref", "x-ref", "primary", "secondary"))
# Local Sequence identifier lcl|identifier
ids.append(make_1id("lcl", "local", "primary"))
# "|" them all together
ncbi_word = Std.dbxref(reduce(operator.or_, ids))
#ncbi_term = Assert(Re("[^ \R]+\|")) + \
ncbi_term = ncbi_word + Rep(Str("|") + ncbi_word)
# Anything else
# BL CCY; width=14; seqs=44; 99.5%=717; strength=1059
BL = Str("BL ") + ToSep("protomat_id", ";") + \
Str(" width=") + Digits("width") + \
Str("; seqs=") + Digits("numseqs") + \
Str("; 99.5%=") + Digits("protomat_count") + \
Str("; strength=") + Digits("strength") + \
AnyEol()
# PLMN_BOVIN|P06868 ( 60) CEEETDFVCRAFQY 26
# ^^^^^^^^^^^^^^^^^
# ^^^^-- number of segments
# ^^^^^^^^^^^^^^-- matching sequence
# ^^-- weight
#
identifier = (Std.dbxref_dbid(UntilSep(sep = "|."),
{"dbname": "swissprot", "type": "primary"}) + \
Str("|") + \
Std.dbxref_dbid(UntilSep(sep = " "),
{"dbname": "swissprot", "type": "accession"})) |\
Std.dbxref_dbid(UntilSep(sep = " "))
segment = AssertNot(Re(r".. ")) + \
identifier + \
Re(r" *\( *") + \
Integer("position") + \
Re(r"\) *") + \
Word("matching_sequence") + Spaces() + \
Digits("weight") + AnyEol()
segment_block = Rep1(segment | AnyEol())
