_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() )
# used in StdHandler for fast dxbref - don't rename!
real_DR_prosite = Std.dbxref(
Std.dbxref_dbname(Martel.Group("database_identifier",
Martel.Str("PROSITE", "PFAM")),
{"style": "sp"}) +
Martel.Str("; ") +
primary_id +
FH_block = Martel.Str("FH Key Location/Qualifiers") + \
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+="))) + \
# 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
generic_term = Std.dbxref(
Std.dbxref_dbid(UntilSep(sep = " "), {"dbname": "local"})
)
id_term = ncbi_term | generic_term
###########################################################
comment_lines = Rep(Str("#") + ToEol())
title = Str(">") + Std.description_line(id_term + UntilEol()) + AnyEol()
seqline = AssertNot(Str(">")) + Std.sequence(UntilEol()) + AnyEol()
# can get a sequence line without an Eol at the end of a file
