make a BED file with exons only

Question

[Giulia.cosenza]

I made a BED file with only exons (hg19) using UCSC table browser (https://genome.ucsc.edu/cgi-bin/hgTables).

But in my file I've found this regions, for example:

chr 1 146310551-146334210

chr 7 9229189-9229487

chr 14 76666574-76669131

chr 17 41341930-41342793

chr 21 19935632-19935690

Do you recognize them as exons? Because they seem to me to be intronic sequences. What can it be the problem?

[Marco Pannone]

What makes you think they are intronic regions? Did you search them into a Genome Browser?

[Giulia.cosenza]

Yes, I used NCBI Variation Viewer.

I asked because maybe I couldn't interpretate well the data on the genome browser.

For example this is what I get for the first interval:

[iraun]

The first interval span is about 23K bp, a bit big for an exon if you ask me. If you want to make a bed file containing the exon coordinates of the human genome, I will recommend you to download the GTF file, extract the exon lines, and convert it to bed format.

[Giulia.cosenza]

Does that file refer to hg19? I'm working with that reference

Answer 1

[Rafael Soler]

You can download the hg19 GTF, for example here: https://www.gencodegenes.org/human/release_19.html

And after this, select only the lines with the word "exon" and extract the coordinates:

grep exon gencode.v19.chr_patch_hapl_scaff.annotation.gtf | cut -f1,4,5 > Human_exons.gtf

Also, you can add "sort -u" to remove duplicate exons

grep exon gencode.v19.chr_patch_hapl_scaff.annotation.gtf | cut -f1,4,5 | sort -u > Human_unique_exons.gtf

Also if you are not working with Linux or a Command Line, you can download them from Biomart https://www.ensembl.org/biomart/martview/0f2873b6b31d9a0aaf1e8f3122bdae0f

I hope it helps! :)

[iraun]

Remember that a BED file is 0-based while a GTF is 1-based coordinate system.

[jena]

True, however note the question is tagged "vcf" so I assume OP wants to filter a VCF file using the BED. In such case OP can use bcftools which will accept a tab-delimited file with 1-based coordinates, such as the one produced by this answer ;) Just make sure it has a different file suffix than .bed (I usually use .tsv or .tab).

As a side note - with many small regions (such as exons), using e.g. bcftools view -T filter.bed is much much faster than bcftools view -R filter.bed, which is better suited for filtering a few larger regions. The -T flag will also let you invert the filter with ^.

[Giulia.cosenza]

Thank you very much!

Answer 2

[elisheva]

You can upload a list of gene ids to the table browser, select the output format as BED and you will see an option to download only exons/introns related to those genes.