Difference between revisions of "Index GFF"
| Line 6: | Line 6: | ||
<ul> | <ul> | ||
<li>Only index feature files that are larger than 5 Mb or even 10 Mb when compressed with GZIP.</li> | <li>Only index feature files that are larger than 5 Mb or even 10 Mb when compressed with GZIP.</li> | ||
| − | <li>Only index feature files that do contain composite features, i.e. feature consisting of multiple locations. The prime example are eukaryote genes. | + | <li>Only index feature files that do contain composite features, i.e. feature consisting of multiple locations. The prime example are eukaryote genes. </li> |
<li>Typically the annotation of a genome does not need to be indexed if it just contains genes, mRNA, CDS and exons</li> | <li>Typically the annotation of a genome does not need to be indexed if it just contains genes, mRNA, CDS and exons</li> | ||
<li>You should not included multiple types (mRNA,CDS, ...) of annotation in a single file as all features will be loaded in a single track with the file name as label. We suggest you put each type in its own file.</li> | <li>You should not included multiple types (mRNA,CDS, ...) of annotation in a single file as all features will be loaded in a single track with the file name as label. We suggest you put each type in its own file.</li> | ||
| Line 26: | Line 26: | ||
== GFF formatted file == | == GFF formatted file == | ||
| + | |||
| + | <b>Warning!: Compound features will be broken up during indexing of gff files.</b> | ||
sort -T /group/tmp -k1,1 -k4,4n input.gff | bgzip -c > compressed.gff.bgz | sort -T /group/tmp -k1,1 -k4,4n input.gff | bgzip -c > compressed.gff.bgz | ||
Revision as of 15:06, 6 September 2013
Large feature files need to be indexed before you can use them properly in GenomeView.
The definition of large is not strict in the sense that it depends on both the real size of the file, as well as the number of features in the file.
Recommendations:
- Only index feature files that are larger than 5 Mb or even 10 Mb when compressed with GZIP.
- Only index feature files that do contain composite features, i.e. feature consisting of multiple locations. The prime example are eukaryote genes.
- Typically the annotation of a genome does not need to be indexed if it just contains genes, mRNA, CDS and exons
- You should not included multiple types (mRNA,CDS, ...) of annotation in a single file as all features will be loaded in a single track with the file name as label. We suggest you put each type in its own file.
Instructions: To index a file, you need to pre-process it with tabix, much like is done with pile-up files.
Tabix can be downloaded from the tabix download page.
BED formatted files
sort -k1,1 -k2,2n input.bed | bgzip -c > compressed.bed.bgz tabix -p bed compressed.bed.bgz
You will get two new files: (1) a bgz file and (2) a tbi file. Load the bgz file in GenomeView.
Note that indexing will not work with BED files that have a UCSC header ("track name=blah")
GFF formatted file
Warning!: Compound features will be broken up during indexing of gff files.
sort -T /group/tmp -k1,1 -k4,4n input.gff | bgzip -c > compressed.gff.bgz tabix -p gff compressed.gff.bgz
You will get two new files: (1) a bgz file and (2) a tbi file. Load the bgz file in GenomeView.
Note that the structure of genes and the type of annotation features will be lost when indexing gff files.