feature = Std.feature(
Martel.Str("FT ") + \
Std.feature_name(Martel.UntilSep(sep = " ")) + \
whitespace + \
Std.feature_location(Martel.UntilEol()) + \
Martel.AnyEol() + \
Martel.Rep(Martel.Str("FT ") + \
Martel.AssertNot(Martel.Str("/")) + \
Std.feature_location(Martel.UntilEol()) + \
Martel.AnyEol()
) + \
Martel.Rep(feature_qualifier)
)
FT_block = Std.feature_block(Martel.Rep(feature),
{"location-style": "genbank"})
## CC - comments or notes (>=0 per entry)
CC_line = Martel.Str("CC ") + \
Martel.ToEol("comment")
CC_block = Martel.Rep1(CC_line)
## XX - spacer line (many per entry)
XX = Martel.Str("XX") + Martel.AnyEol()
## SQ - sequence header (1 per entry)
SQ_line = Martel.Str("SQ Sequence ") + \
Martel.Digits("num_BP") + \
qualifier = Std.feature_qualifier(
qualifier_space +
Martel.Str("/") +
Std.feature_qualifier_name(Martel.Alt(*feature_qualifiers)) +
(Martel.AnyEol() | # '/pseudo'
(Martel.Str("=") +
(unquoted_string | # '/evidence=experimental'
quoted_string))) # '/translation="AAAAAAAA....
# AAAAAAAAAAAAAAAAAAAA'
)
feature = Std.feature(feature_key_line +
Martel.Rep(qualifier))
feature_block = Std.feature_block(
features_line + \
Martel.Rep1(feature),
{"location-style": "genbank"}
)
# BASE COUNT 28300 a 15069 c 15360 g 27707 t
base_count = Martel.Group("base_count",
Martel.Re("[\w\d ]+"))
base_count_line = Martel.Group("base_count_line",
Martel.Str("BASE COUNT") +
blank_space +
base_count +
Martel.AnyEol())
# ORIGIN
# 1 ggacaaggcc aaggatgctg ctgctgcagc tggagcttcc gcgcaacaag taaacagata
Martel.Str(" ") + \
FT_end + \
Martel.Opt(Martel.Str(" ") + \
FT_desc) + \
Martel.AnyEol()
FT_continuation = Martel.Str("FT ") + \
FT_desc + \
Martel.AnyEol()
FT = Std.feature(FT_range + Martel.Rep(FT_continuation),
{"location-style": "sp"})
##feature_block = Martel.Group("feature_block", Martel.Rep1(FT))
feature_block = Std.feature_block(Martel.Rep1(FT),
{"style": "swissprot"})
#--- SQ
# SQ SEQUENCE XXXX AA; XXXXX MW; XXXXX CRC32;
# (Those X's don't really indicate the size)
SQ = Martel.Group("SQ",
Martel.Re("SQ SEQUENCE +(?P<sequence_length>\d+) AA;" \
" +(?P<molecular_weight>\d+) MW;" \
" +(?P<crc?type=32>\w+) CRC32;\R")
)
##SQ_data = Martel.Group("SQ_data",
## Martel.Re(" (?P<sequence>[^\R]*)\R"))
SQ_data = Martel.Str(" ") + \
