Skip to content

Commit

Permalink
Notify errors for standard tags in the INFO column from a VCF record. #…
Browse files Browse the repository at this point in the history 
…18
  • Loading branch information
Cristina Yenyxe Gonzalez Garcia committed Jan 8, 2015
1 parent fe29e63 commit 2dbb774
Show file tree
Hide file tree
Showing 3 changed files with 3,073 additions and 2,071 deletions.
182 changes: 147 additions & 35 deletions cpp/src/bioformats/vcf/vcf_v41.ragel
View file
Original file line number Diff line number Diff line change
Expand Up @@ -205,7 +205,7 @@
}


########## Incorrect fields actions ##########
########## Incorrect metadata and header actions ##########

# Fileformat line
action fileformat_error {
Expand Down Expand Up @@ -339,48 +339,151 @@
fhold; fgoto meta_section_skip;
}

# Records

########## Incorrect records actions ##########

# Chromosome
action chrom_error {
ErrorPolicy::handle_body_section_error(*this, "Chromosome is not a string without colons or whitespaces, optionally wrapped with angle brackets (<>)");
fhold; fgoto body_section_skip;
}

# Position
action pos_error {
ErrorPolicy::handle_body_section_error(*this, "Position is not a positive number");
fhold; fgoto body_section_skip;
}

# ID
action id_error {
ErrorPolicy::handle_body_section_error(*this, "ID is not a single dot or a list of strings without semicolons or whitespaces");
fhold; fgoto body_section_skip;
}

# Reference allele
action ref_error {
ErrorPolicy::handle_body_section_error(*this, "Reference is not a string of bases");
fhold; fgoto body_section_skip;
}

# Alternate alleles
action alt_error {
ErrorPolicy::handle_body_section_error(*this, "Alternate is not a single dot or a comma-separated list of bases");
fhold; fgoto body_section_skip;
}

# Quality
action qual_error {
ErrorPolicy::handle_body_section_error(*this, "Quality is not a single dot or a positive number");
fhold; fgoto body_section_skip;
}

# Filter
action filter_error {
ErrorPolicy::handle_body_section_error(*this, "Filter is not a single dot or a semicolon-separated list of strings");
fhold; fgoto body_section_skip;
}

# Info
action info_error {
printf("Line %zu: Error in 'info' field\n", n_lines);
ErrorPolicy::handle_body_section_error(*this, "Info is not a single dot or a semicolon-separated list of key-value pairs");
fhold; fgoto body_section_skip;
}

#info_key = (alnum | (punct - (";" | "=")))+ - ( "AA" | "AC" | "AF" | "AN" | "BQ" | "CIGAR" | "DB" | "DP" | "END" | "H2" | "H3" | "MQ" | "MQ0" | "NS" | "SB" | "SOMATIC" | "VALIDATED" | "1000G" | (punct)+ ) ;
#info_value = (print - space)+ ;
#info_value_list = info_value ("," info_value)* ;

action info_AA_error {
ErrorPolicy::handle_body_section_error(*this, "Info AA value is not a single dot or a string of bases");
fhold; fgoto body_section_skip;
}

action info_AC_error {
ErrorPolicy::handle_body_section_error(*this, "Info AC value is not a comma-separated list of numbers");
fhold; fgoto body_section_skip;
}

action info_AF_error {
ErrorPolicy::handle_body_section_error(*this, "Info AF value is not a comma-separated list of numbers");
fhold; fgoto body_section_skip;
}

action info_AN_error {
ErrorPolicy::handle_body_section_error(*this, "Info AN value is not an integer number");
fhold; fgoto body_section_skip;
}

action info_BQ_error {
ErrorPolicy::handle_body_section_error(*this, "Info BQ value is not a number");
fhold; fgoto body_section_skip;
}

action info_CIGAR_error {
ErrorPolicy::handle_body_section_error(*this, "Info CIGAR value is not an alphanumeric string");
fhold; fgoto body_section_skip;
}

action info_DB_error {
ErrorPolicy::handle_body_section_error(*this, "Info DB is not a flag (with 1/0/no value)");
fhold; fgoto body_section_skip;
}

action info_DP_error {
ErrorPolicy::handle_body_section_error(*this, "Info DP value is not an integer number");
fhold; fgoto body_section_skip;
}

action info_END_error {
ErrorPolicy::handle_body_section_error(*this, "Info END value is not an integer number");
fhold; fgoto body_section_skip;
}

action info_H2_error {
ErrorPolicy::handle_body_section_error(*this, "Info H2 is not a flag (with 1/0/no value)");
fhold; fgoto body_section_skip;
}

action info_H3_error {
ErrorPolicy::handle_body_section_error(*this, "Info H3 is not a flag (with 1/0/no value)");
fhold; fgoto body_section_skip;
}

action info_MQ_error {
ErrorPolicy::handle_body_section_error(*this, "Info MQ value is not a number");
fhold; fgoto body_section_skip;
}

action info_MQ0_error {
ErrorPolicy::handle_body_section_error(*this, "Info MQ0 value is not an integer number");
fhold; fgoto body_section_skip;
}

action info_NS_error {
ErrorPolicy::handle_body_section_error(*this, "Info NS value is not an integer number");
fhold; fgoto body_section_skip;
}

action info_SB_error {
ErrorPolicy::handle_body_section_error(*this, "Info SB value is not a number");
fhold; fgoto body_section_skip;
}

action info_SOMATIC_error {
ErrorPolicy::handle_body_section_error(*this, "Info SOMATIC is not a flag (with 1/0/no value)");
fhold; fgoto body_section_skip;
}

action info_VALIDATED_error {
ErrorPolicy::handle_body_section_error(*this, "Info VALIDATED is not a flag (with 1/0/no value)");
fhold; fgoto body_section_skip;
}

action info_1000G_error {
ErrorPolicy::handle_body_section_error(*this, "Info 1000G is not a flag (with 1/0/no value)");
fhold; fgoto body_section_skip;
}

# Format
action format_error {
ErrorPolicy::handle_body_section_error(*this, "Format is not a colon-separated list of alphanumeric strings");
fhold; fgoto body_section_skip;
Expand All @@ -391,6 +494,7 @@
fhold; fgoto body_section_skip;
}

# Samples
action sample_error {
std::ostringstream message_stream;
message_stream << "Sample #" << (n_columns - 9) << " is not a valid string";
Expand Down Expand Up @@ -445,24 +549,32 @@
( ":" meta_field_value)? $err(meta_alt_id_suffix_err)
)
",Description=\"" meta_field_desc "\"" $err(meta_desc_err) ;

meta_assembly = url $err(meta_url_err) ;

## TODO Probably need to check meta_contig again after getting an answer from the spec team
meta_contig = "ID=" identifier $err(meta_id_err)
",URL=" url $err(meta_url_err) ;

meta_filter = "ID=" identifier $err(meta_id_err)
",Description=\"" meta_field_desc "\"" $err(meta_desc_err) ;

meta_format = "ID=" identifier $err(meta_id_err)
",Number=" ( (digit)+ | "A" | "R" | "G" | "." ) $err(meta_format_number_err)
",Type=" ( "Integer" | "Float" | "Character" | "String" ) $err(meta_format_type_err)
",Description=\"" meta_field_desc "\"" $err(meta_desc_err) ;

meta_info = "ID=" identifier $err(meta_id_err)
",Number=" ( (digit)+ | "A" | "R" | "G" | "." ) $err(meta_info_number_err)
",Type=" ( "Integer" | "Float" | "Flag" | "Character" | "String" ) $err(meta_info_type_err)
",Description=\"" meta_field_desc "\"" $err(meta_desc_err)
(",Source=\"" meta_field_desc "\"")? $err(meta_info_source_err)
(",Version=\"" meta_field_desc "\"")? $err(meta_info_version_err) ;

meta_pedigree = identifier $err(meta_id_err) "=" identifier $err(meta_id_err) ("," identifier $err(meta_id_err) "=" identifier $err(meta_id_err))* ;

meta_pedigreeDB = url $err(meta_url_err) ;

meta_sample = "ID=" identifier $err(meta_id_err)
(",Genomes=" identifier (";" identifier)* ) $err(meta_sample_genomes_err)
(",Mixture=" identifier (";" identifier)* ) $err(meta_sample_mixture_err)
Expand Down Expand Up @@ -530,32 +642,32 @@
filter_value = (alnum | punct - ";")+ - (punct)+ ;
record_filter = (filter_value (";" filter_value)*) | "." ;

info_key = ( (alnum | "_")+ | "." ) - ("AA" | "AC" | "AF" | "AN" | "BQ" | "CIGAR" | "DB" | "DP" | "END" | "H2" | "H3" | "MQ" | "MQ0" | "NS" | "SB" | "SOMATIC" | "VALIDATED" | "1000G") ;
info_value = (print - space)+ ;
info_key = (alnum | (punct - (";" | "=")))+ - ( "AA" | "AC" | "AF" | "AN" | "BQ" | "CIGAR" | "DB" | "DP" | "END" | "H2" | "H3" | "MQ" | "MQ0" | "NS" | "SB" | "SOMATIC" | "VALIDATED" | "1000G" | (punct)+ ) ;
info_value = (print - (space | ";"))+ ;
info_value_list = info_value ("," info_value)* ;
info_entry = (
"AA=" (bases | ".") |
"AC=" int_number ("," int_number)* |
"AF=" any_number ("," any_number)* |
"AN=" int_number |
"BQ=" any_number |
"CIGAR=" (alnum)+ |
"DB" ("=" ("1" | "0"))? |
"DP=" int_number |
"END=" int_number |
"H2" ("=" ("1" | "0"))? |
"H3" ("=" ("1" | "0"))? |
"MQ=" any_number |
"MQ0=" int_number |
"NS=" int_number |
"SB=" any_number |
"SOMATIC" ("=" ("1" | "0"))? |
"VALIDATED" ("=" ("1" | "0"))? |
"1000G" ("=" ("1" | "0"))? |
"AA=" (bases | ".") $err(info_AA_error) |
"AC=" (int_number ("," int_number)*) $err(info_AC_error) |
"AF=" (any_number ("," any_number)*) $err(info_AF_error) |
"AN=" int_number $err(info_AN_error) |
"BQ=" any_number $err(info_BQ_error) |
"CIGAR=" (alnum)+ $err(info_CIGAR_error) |
"DB" ("=" ("1" | "0"))? $err(info_DB_error) |
"DP=" int_number $err(info_DP_error) |
"END=" int_number $err(info_END_error) |
"H2" ("=" ("1" | "0"))? $err(info_H2_error) |
"H3" ("=" ("1" | "0"))? $err(info_H3_error) |
"MQ=" any_number $err(info_MQ_error) |
"MQ0=" int_number $err(info_MQ0_error) |
"NS=" int_number $err(info_NS_error) |
"SB=" any_number $err(info_SB_error) |
"SOMATIC" ("=" ("1" | "0"))? $err(info_SOMATIC_error) |
"VALIDATED" ("=" ("1" | "0"))? $err(info_VALIDATED_error) |
"1000G" ("=" ("1" | "0"))? $err(info_1000G_error) |
info_key |
info_key "=" info_value_list
);
record_info = info_entry (";" info_entry)* ;
record_info = info_entry (";" info_entry)* | "." ;

## TODO Could this be extended because a lot of files use _ as a separator?
format_value = (alnum)+ ;
Expand All @@ -566,16 +678,16 @@
sample_gt = sample_allele (("/" | "|") sample_allele)* ;
record_sample = sample_gt $err(sample_gt_error) (":" (alnum | punct)+)* ;

record = (record_chrom >token_begin %chrom_end $err(chrom_error)
CS record_pos >token_begin %pos_end $err(pos_error)
CS record_id >token_begin %id_end $err(id_error)
CS record_ref >token_begin %ref_end $err(ref_error)
CS record_alt >token_begin %alt_end $err(alt_error)
CS record_qual >token_begin %qual_end $err(qual_error)
CS record_filter >token_begin %filter_end $err(filter_error)
CS record_info >token_begin %info_end $err(info_error)
(CS record_format >token_begin %format_end $err(format_error)
(CS record_sample >token_begin %sample_end $err(sample_error))+ )?
record = (record_chrom >token_begin %chrom_end $err(chrom_error)
CS record_pos >token_begin %pos_end $err(pos_error)
CS record_id >token_begin %id_end $err(id_error)
CS record_ref >token_begin %ref_end $err(ref_error)
CS record_alt >token_begin %alt_end $err(alt_error)
CS record_qual >token_begin %qual_end $err(qual_error)
CS record_filter >token_begin %filter_end $err(filter_error)
CS record_info >token_begin %info_end $err(info_error)
(CS record_format >token_begin %format_end $err(format_error)
(CS record_sample >token_begin %sample_end $err(sample_error))+ )?
) >record_begin %record_end;

fileformat_section = (fileformat NL) $err(fileformat_section_error);
Expand Down
Loading

0 comments on commit 2dbb774

Please sign in to comment.