build 37 LOKI

[katiecardone26]

Is your feature request related to a problem? Please describe.
Biofilter is compatible with both b37 and b38 genomes. However, the b37 version of LOKI on the LPC is from 2014, whereas the b38 version is from 2022. I think there should be a more up to date version of b37 LOKI.

Describe the solution you'd like
I think there should be a more up to date b37 version of LOKI. I think this could be achieved by using liftover to convert the b38 version to b37.

[AndreRico]

Hi @katiecardone26 , would you like to discuss this point in our next topic meeting? We would implement it in version 2.4.3?

[katiecardone26]

I think it would be helpful to implement in version 2.4.3, because I believe b37 annotations can be queried with the existing databases. I don't think I am on the biofilter/loki project meetings but feel free to bring up this point at the next meeting.

[AndreRico]

Hi @katiecardone26!

The latest version of the LOKI database is built on GRCh38, while I understand that you intend to provide positions from a different assembly (e.g., GRCh37). Is this correct?

To address this, I created a test scenario where a different assembly is provided as input than the one the LOKI data file was built on. I observed that Biofilter compares the assembly version of the LOKI file with the one specified in the arguments (--ucsc-build-version, e.g., "19"). If a mismatch is detected, a conversion table (e.g., from GRCh37 to GRCh38) is used to convert the input positions, and all queries are processed using the converted coordinates.

The output will retain the original positions as labels (e.g., GRCh37/hg19), while the results will be aligned with the GRCh38 coordinates.

I have documented this process in more detail here: (https://github.com/RitchieLab/biofilter/tree/development/tests/issues/l16_build_37_loki)

Does this approach address your requirements? If not, I’d be happy to discuss alternative solutions.