WormBase is released in a number of data formats. The GFF files associated with each release constitute a simple flatfile version of the sequence and sequence features. This document explains the naming of features and methods utilised in the GFF files.

GFF format

Briefly, a GFF file is a tab-delimited flatfile with the following structure:

‹seqname› ‹source› ‹feature› ‹start› ‹end› ‹score› ‹strand› ‹frame› ‹attributes›

see GFF specification for a fuller explanation of the GFF format.

Current GFF Source list

Source	Possible features	Description
Genomic_canonical	Sequence	Genome sequences forming the minimal tiling path along each chromosome. The named sequences represent those objects in ACEDB declared 'Genomic_canonical' and these are the sequences submitted to the public nucleotide databases. The full list of clone paths can be found in the summary tables for each WormBase release.
Link	Sequence	Level of organisation of the genome sequences into chromosome sequences. Clones are aligned into LINK and SUPERLINK objects to form chromosomes.
curated	Sequence exon intron CDS	Gene prediction for a protein-coding sequence. 'Curated' genes have been appraised by an annotator. The nomenclature for curated CDS's is ‹Clone›.[1-99]. Curated predictions are the primary data for generating the wormpep data set.
provisional	Sequence exon intron CDS	Gene prediction for a protein-coding sequence. 'Provisional' genes are preliminary Genefinder predictions and have not been appraised by an annotator. The nomenclature for provisional CDS is ‹Clone›.[a-z]. 'provisional' predictions are included in the wormpep data set.
tRNAscan-SE-1.11	Sequence exon intron	tRNA gene predictions. tRNAs are predicted using the tRNAscan program (Sean Eddy).
RNA	Sequence exon intron	RNA gene predictions.
Pseudogene	Sequence exon intron	Pseudogene predictions.
Transposon	repeat	Transposon predictions.
GenePair_STS	Structural	The PCR product which results from the amplification using a defined set of oligonucleotides. The PCR_products constitute a resource for investigating C.elegans biology especially functional analysis. Hence, a number of features have utilised these sequences and the associated data will have identical coordinates (e.g. RNAi phenotypes and Expression profile analysis).
WTP	partial_gene	Worm Transcriptome Project (WTP) predictions of partial trasncripts. Essentially mapping the EST/mRNA transcript data to genomic sequence and marking the extent of overlap. WTP spans relate to confirmed transcripts and form a data set of experimentally supported worm genes. The exon/intron sructure is not stored and the span includes potential 5' and 3' UTR sequences.
RNAi	experimental	RNAi experiments. The start/stop span relates to the physical DNA used in the RNAi assay. RNAi has been performed using PCR products and cDNA clones.
cDNA_for_RNAi	experimental	cDNA sequences used in RNAi assays. The start/stop span relates to the full length mapping of the cDNA to genomic sequence (i.e. it includes the introns of the unprocessed transcript).
Expr_profile	Expression	Stuart Kym's expression profiles. A distillation of microarray expression studies based on the Research Genetics designed oligo data set. Each profile relates to the PCR product on the microarray slide.
BLAT_EST_BEST	similarity	BLAT alignment of EST sequence to genomic sequence. The best alignment indicates this is the optimal alignment in the current release.
BLAT_EST_OTHER	similarity	BLAT alignment of EST sequence to genomic sequence. The other alignment indicates this is a sub-optimal alignment in the current release and a better match is found elsewhere.
BLAT_mRNA_BEST	similarity	BLAT alignment of mRNA sequence to genomic sequence. The best alignment indicates this is the optimal alignment in the current release.
BLAT_mRNA_OTHER	similarity	BLAT alignment of mRNA sequence to genomic sequence. The other alignment indicates this is a sub-optimal alignment in the current release and a better match is found elsewhere.
BLATX_NEMATODE	similarity	BLATX alignment of nematode EST consensus sequences to genomic sequence.
WABA_weak	similarity
WABA_strong	similarity
WABA_coding	similarity
BLASTX	similarity
inverted	repeat	Inverted repeat regions.
tandem	repeat	Tandem repeat regions.
RepeatMasker	repeat	Dispersed repeats as mapped by RepeatMasker
Allele	SNP complex_change_in_nucleotide_sequence deletion insertion substitution transposable_element_insertion_site	Known differences from the reference genome sequence.

Current GFF Feature list

Feature	Quoted in	Description
Sequence	Genomic_canonical Link curated provisional Pseudogene tRNAscan RNA	Sequence features relate to regions defined by a start and stop coordinate. Sequences have a direction based on the strand attribute. See Figure 1 feature A for an example of the sequence feature for a gene prediction.
exon	curated provisional Pseudogene tRNAscan RNA	Exons features are part of predicted gene models. An exon is defined by a start and stop coordinate with a direction based on the strand attribute. See Figure 1 feature B for an example of the sequence feature for a protein-coding gene prediction. Exon do not necessarily correlate with coding sequences (see exon 1 and exon 4 in the Figure 1 example) - hence an exon feature will not have a frame(phase) attribute.
intron	curated provisional Pseudogene tRNAscan RNA	Intron features are part of predicted gene models. An intron is defined by a start and stop coordinate with a direction based on the strand attribute. See Figure 1 feature D for an example of the sequence feature for a protein-coding gene prediction. Introns do not form part of coding sequence.
CDS	curated provisional	CDS features are part of predicted gene models. An CDS feature is defined by a start and stop coordinate with a direction based on the strand attribute. See Figure 1 feature C for an example of the sequence feature for a protein-coding gene prediction. CDS regions are those which can be translated into peptide sequence. CDS sequences must correlate with coding sequences and hence must have a frame(phase) attribute.
structural	Genepair_STS	Genomic DNA region representing a physical DNA substrate.
partial_gene	WTP	Sequence region which covers a partial gene prediction. This span has no exon/intron structure information. The region is supported by transcript data (essentially EST sequences) but the full span may not translate into a valid peptide.
experimental	RNAi cDNA_for_RNAi	waffle
expression	Expr_profile	Expression studies.
similarity	BLAT_EST_BEST BLAT_EST_OTHER BLAT_mRNA_BEST BLAT_mRNA_OTHER BLATX_NEMATODE WABA_weak WABA_strong WABA_coding BLASTX	Similarity matches.
repeat	transposon inverted tandem RepeatMasker	Repeat features.
ALLELE		no data
substitution	Allele	Simple allelic changes, usually mono or di-nucleotide substitutions
complex_change_in_nucleotide_sequence	Allele	Basically, these are deletion alleles where the deletion is replaced by (a usually smaller length) insertion sequence
insertion	Allele	Simple insertion alleles, which are not due to transposon insertions. Allele object should (ideally) capture nature of inserted sequence
deletion	Allele	Deletion alleles in WormBase represent where other strains of C. elegans have a tract of genomic sequence deleted (with respect to the sequenced N2 strain). Many deletion alleles in WormBase were generated by the C. elegans Knockout Consortium, though details of many other deletion alleles have been extracted from the literature. Some deletion alleles represent deletions which have also been associated with a insertion, i.e. a large tract of sequence is deleted and replaced by a smaller tract of different sequence.
	Allele
Clone_left_end
Clone_right_end

Figure 1 - Example of a sequence region which contains a gene model. Exons are shown as black boxes linked by introns. The coding sequence (ATG->STOP) is shaded in red. The spans (coordinate start/stop) for GFF features are marked as horizontal coloured bars. Features displayed include [A] Sequence, [B] exon, [C] CDS coding exon and [D] introns.