Skip to content

Commit

Permalink
Merge pull request #9 from stjudecloud/updated-file-types
Browse files Browse the repository at this point in the history 
feat: updated file types
  • Loading branch information
@jakenoble
jakenoble authored Jan 7, 2025
2 parents 95e565a + f4052ec commit 82d3062
Showing 1 changed file with 114 additions and 8 deletions.
122 changes: 114 additions & 8 deletions src/CDE/V1/Data/FileType.php
View file
Original file line number Diff line number Diff line change
Expand Up @@ -78,6 +78,16 @@ enum FileType implements ArrayAccess
case XML;
case YAML;
case ZIP;
case CNS;
case DICT;
case DOC;
case RTF;
case SEG;
case TBI;
case CDF;
case IDPDB;
case RDS;
case PED;

const CDE_ID = 11416926;

Expand Down Expand Up @@ -360,10 +370,9 @@ enum FileType implements ArrayAccess
'public_id' => 14818706,
'concept_code' => 'C204352',
'description' => 'A text file format used for storing gene sequence variations. It is a type
of variant call format that has a record for all sites, whether there is a
variant call there or not. The format includes metrics of the confidence
that positions are actually non-variant vs. a factor of minimum read-depth
and genotype quality.',
of variant call format that has a record for all sites, whether there is a variant call there
or not. The format includes metrics of the confidence that positions are actually non-variant
vs. a factor of minimum read-depth and genotype quality.',
],
'GZIP Format' => [
'value' => self::GZIP_FORMAT,
Expand Down Expand Up @@ -409,9 +418,13 @@ enum FileType implements ArrayAccess
'value' => self::HTSEQ,
'permissible_value' => 'HTSeq Count',
'long_name' => 'HTSeq-Count Output Format',
'public_id' => 11422133,
'concept_code' => 'C184763',
'description' => 'A tab-delimited file generated by the HTSeq Count software tool. The file represents the number of reads obtained during high-throughput sequencing that map to known genomic sequences. The data is represented in two columns, one containing an identifier for a genomic feature, which can be a database-supplied identifier or a gene symbol, followed by a second column containing the number of mapped reads.',
'public_id' => 14943335,
'concept_code' => 'C205749',
'description' => 'A tab-delimited file generated by the HTSeq Count software tool. The
file represents the number of reads obtained during high-throughput sequencing that map
to known genomic sequences. The data is represented in two columns, one containing an
identifier for a genomic feature, which can be a database-supplied identifier or a gene
symbol, followed by a second column containing the number of mapped reads.',
],
'IDAT' => [
'value' => self::IDAT,
Expand Down Expand Up @@ -770,5 +783,98 @@ enum FileType implements ArrayAccess
'description' => 'An archive file format that supports lossless data compression. It is used
to compress one or more files together into a single location.',
],
'CNS' => [
'value' => self::CNS,
'permissible_value' => 'CNS',
'long_name' => 'Cytoscape Input File Format',
'public_id' => 15354253,
'concept_code' => 'C209889',
'description' => 'A gene expression report format that utilizes a graph with genes,
proteins, and molecules represented as node and their interactions represented as
links.',
],
'DICT' => [
'value' => self::DICT,
'permissible_value' => 'DICT',
'long_name' => 'HET Group Dictionary Entry Format',
'public_id' => 15354254,
'concept_code' => 'C209890',
'description' => 'The simple flat file format of an entry from the HET group dictionary,
a PDB file resource for ligand groups that are found complexed to protein structures.',
],
'DOC' => [
'value' => self::DOC,
'permissible_value' => 'DOC',
'long_name' => 'DOC Format',
'public_id' => 15354330,
'concept_code' => 'C210442',
'description' => 'A Microsoft proprietary Binary Interchange file format used for word
processing documents. It was replaced with DOCX format in 2007.',
],
'RTF' => [
'value' => self::RTF,
'permissible_value' => 'RTF',
'long_name' => 'Rich Text Format',
'public_id' => 15092264,
'concept_code' => 'C209896',
'description' => 'A file format that contains various formatting elements and enables
exchanges of text files across different word processing platforms and applications.',
],
'SEG' => [
'value' => self::SEG,
'permissible_value' => 'SEG',
'long_name' => 'Segmented Data Format',
'public_id' => 15092263,
'concept_code' => 'C209892',
'description' => 'A tab-delimited text file that is the tabular output of the Circular
Binary Segmentation algorithm implemented in DNAcopy and lists loci and associated
numeric values.',
],
'TBI' => [
'value' => self::TBI,
'permissible_value' => 'TBI',
'long_name' => 'TBI Format',
'public_id' => 15042846,
'concept_code' => 'C184806',
'description' => 'A tab-delimited file that contains an index for the genome positions
in a compressed (zipped) variant call format file.',
],
'CDF' => [
'value' => self::CDF,
'permissible_value' => 'cdf',
'long_name' => 'Affymetrix Probe Sets Library File',
'public_id' => 15354255,
'concept_code' => 'C209894',
'description' => 'A file that provides information about the probes contained
within an Affymetrix expression microarray probe set.',
],
'IDPDB' => [
'value' => self::IDPDB,
'permissible_value' => 'idpDB',
'long_name' => 'IDPicker File Format',
'public_id' => 15354267,
'concept_code' => 'C209893',
'description' => 'A self-contained file format that utilizes an open-source protein
assembly tool to view and filter peptide-spectrum-matches, and stores relationships
between proteins, peptides, and spectra.',
],
'RDS' => [
'value' => self::RDS,
'permissible_value' => 'rds',
'long_name' => 'Reference Data Set File Format',
'public_id' => 15354268,
'concept_code' => 'C209895',
'description' => 'A binary file format that provides a method for saving R data
sets and other objects.',
],
'PED' => [
'value' => self::PED,
'permissible_value' => 'PED',
'long_name' => 'Pedigree File Format',
'public_id' => 15092262,
'concept_code' => 'C209891',
'description' => 'A standard text format for sample pedigree information and storage
of marker genotypes, disease status, and quantitative trait values.',
],
];
}
}

0 comments on commit 82d3062

Please sign in to comment.