Extend bed formats to bed classifier, e.g. gappedPeak

[donaldcampbelljr]

We should add more bed formats, e.g.
https://genome.ucsc.edu/FAQ/FAQformat.html

[donaldcampbelljr]

We should also create an ontology of genomic region file formats in JSON format to be hosted on PEPhub:
narrowPeak + broadPeak vs bed 6+4 etc

[donaldcampbelljr]

gappedPeak has now been added to the classifier

[donaldcampbelljr]

I investigated historical formats this morning.

Do we want to add these to bedbase?

ENCODE tagAlign: BED3+3 format (historical)

• This format used hg18.
• GEO only had ~ 142 files, interestingly none of them followed the spec where column 4 is a reported sequence, instead all files had 'N' in column 4.
• ENCODE had over 5000 results, many if not all were labeled as archived.

ENCODE pairedTagAlign: BED6+2 format (historical)

• None found in GEO or ENCODE when searching

ENCODE peptideMapping: BED6+4 format

• These simply have a .bed extension
• No hits on GEO, only 68 hits on ENCODE
• They are similar to narrowPeaks but we could possibly use the 9th column to help differentiate between the two (int vs float).

[donaldcampbelljr]

No to TagAlign.
No to pairedTagAlign.

The above two are actually reads.

No to peptideMapping -> appears to be historical, only a small set of data available.

[donaldcampbelljr]

Investigating ENCODE RNA elements: BED6 + 3 scores format
https://genome.ucsc.edu/FAQ/FAQformat.html#format11

There are no examples currently uploaded to bedbase out of the 1500 bed6+3's. After some internet research, I did find an example here.

I added some logic with the above commit which checks to see if the 9th column is an integer value (but NOT -1).

Example test file:

chr1	56969	57098	id-1	19	.	-1	-1	10
chr1	180739	180871	id-2	144	.	-1	-1	20
chr1	181108	181267	id-3	76	.	-1	-1	30

vs a broadpeak:

DEBUG: COLUMN TYPE VALUE 6 float64 3.4429 7 float64 5.7737 8 float64 4.2324
('bed6+3', 'broadpeak')
DEBUG: COLUMN TYPE VALUE 6 int64 -1 7 int64 -1 8 int64 10
('bed6+3', 'encode_rna_elements')