Problems reading from bed files in Table Browser.

Question

[HumeMarx]

Hi All

I am trying to upload a list of positions to table browser so that I can get their related SNPs. When I try to look for each position using the "position" option in the region section of the Table Browser I get the appropriate SNP. But when I try to do the same using tab delimited files saved as .bed files I get nothing. Even when I paste in to the regions box, I get nothing # No results in given region.

Does anyone know a reason for this?

I can do it one by one, but I don't really fancy doing this 89,000 times. So a solution would be most useful.

Thanks

[HumeMarx]

Command lines scare me, but will do if absolutely unavoidable.

Do you know why the normal pasting in to

identifiers (names/accessions) field to get positions or

"Define region" to get rs lists for a list of positions doesn't work?

[Alex Reynolds]

I don't know. You might double-check the format of your input files. Are they tab-delimited? Did you get them out of an Excel spreadsheet, which means they will contain non-standard line delimiters? A lot of these things can be checked but may require some command-line work.

[HumeMarx]

Thanks. It is a formatting issue as I know the SNPs do exist at these positions. I just can't figure out what.

What would you recommend if I don't use excel? Or do you know where I can look to rectify the problem?

Answer 1

[Alex Reynolds]

If you can do things on the command line, you could use BEDOPS bedmap.

First, sort your positions file with sort-bed:

$ sort-bed unsorted-positions.bed > positions.bed

Second, download SNP data from UCSC to a BED file. For example, if you are working with human hg19 data:

$ mysql --user=genome --host=genome-mysql.cse.ucsc.edu -A -D hg19 -e 'SELECT chrom, chromStart, chromEnd, name FROM snp138Common' | tail -n +2 > snp138Common.bed

Third, map the SNPs to each position with bedmap:

$ bedmap --echo --echo-map-id-uniq positions.bed snp138Common.bed > answer.bed

The file answer.bed contains each position on its own line, along with the rs* IDs that overlap each position.

[HumeMarx]

Thank you for this.

I think I may end up doing this anyway as I am trying to get rid of my irrational fear of commands.