map locations in GrCh38 to GrCh37/hg19 (yes, in reverse)

[SrCardgage]

I have downloaded several restriction enzyme cut sites (single bp location) that have been mapped on the newest human genome build, GrCh38. However, ALL of my previous analysis is on GrCh37/hg19.

There are web pages that map FORWARD to GrCh38. I need to go BACKWARD. My google-fu has failed to find this.

Any help is appreciated.

[GenoMax]

You will find the liftover file you need here: http://hgdownload.soe.ucsc.edu/golde...hg38/liftOver/

[SrCardgage]

Wonderful! Thank you.

[SrCardgage]

Ugh. It's not working.

I downloaded the file hg38ToHg19.over.chain.gz and downloaded the latest build of the "liftOver" program.

I ran it (more or less) as:

liftOver grch38_location.bed hg38ToHg19.over.chain.gz hg19_location.bed err.log

But the log file (which holds unmapped entries) says EVERY entry has been removed. Specifically, each reading has the message:

#Deleted in new

The only oddity about my input files is that every entry is one base pair. So, an example BED entry would be:

chr1 12411 12411

with tabs as delimiters

Do I need to alter these files to add one to each second position? Any other ideas?

Thanks for any help.

~EDIT~
I saw a post about converting SNPs and this person's BED file had the equivalent of:
chr1 12411 12412
I'm writing a script to alter these files. So, if you know this is the answer, no need to reply. I will update with my results.

[GenoMax]

see: https://groups.google.com/a/soe.ucsc...me/X1AhPz8Ozkc That probably applies to the command line liftover as well.

chr1 12411 14412

[SrCardgage]

Success! I did indeed need a "width" of one, not zero, for the input BED files.

Results:
a) 2,317,719 total entries, only 28,414 (1.2%) did not map
b) 7,199,381 total entries, only 118,690 (1.6%) did not map

I consider this to be very acceptable.

Thank you, GenoMax for your quick responses!

[SrCardgage]

I have found duplicate entries in the file that "liftOver" produces. I'm guessing some cut sites were close together and were therefore "joined".

I think the best way to remove these is:
sort -u liftOver_output.txt > unique_output.txt

Don't use "uniq" as it appears to not deal with large files very well. Learned a new unix lesson, today!