Hi,
I am trying to run the FineMapping_AlzheimerDisease notebook in gentropy repo but I’m encountering a permission error when the code attempts to access GWAS Summary Statistics from Google Cloud Storage. Particularly, the issue is from this code
path_gwas1=“gs://gwas_catalog_data/harmonised_summary_statistics/GCST90012877.parquet”
path_si=“gs://gwas_catalog_data/study_index”
gwas1 = SummaryStatistics.from_parquet(session, path_gwas1)
study_index = StudyIndex.from_parquet(session, path_si)
slt=WindowBasedClumping.clump(gwas1,gwas_significance=5e-8,distance=1e6)
slt_df=slt._df
When running the above code, I get:
does not have storage.objects.get access to the Google Cloud Storage object. Permission ‘storage.objects.get’ denied on resource (or it may not exist).", “reason”: “forbidden”
Where can I find the GCST90012877 study file? I have setup gcloud locally using the suggested tool in cell 1 (.ie !gcloud auth application-default login)
Hi @Abdu_Omar and welcome to the Open Targets Community! 
The notebook you are using contains outdated code that we are not maintaining or providing support for, and we cannot provide the input file that you are trying to access. We apologise for this!
We will delete the notebook from the repository to avoid confusion.
Could you give us some more information and context for what you are trying to do? Maybe we can provide some help or feedback to help you.
Hi,
Thanks for your response. I’d like to thank the OpenTargets team for providing their data and code for the broader research community!
Just to check, is it only the code in this particular notebook that is outdated or all of the gentropy codebase is outdated? I thought it was the latest version for opentargets fine-mapping pipeline.
I’m trying to reproduce the results of the Locus-to-Gene (L2G) model introduced in Mountjoy et. al 2021, which I believe is used by the current OpenTargets platform for gene prioritization. Generally, I’m conducting research in building models for causal gene prediction and looking to compare my results with the ones reported in the paper using the Gold Standard Benchmark. If there is a better way of doing this, please let me know
Hi! Thank you for your interest in using Gentropy for your research. The code base is up to date and is constantly updated. However, we haven’t updated the notebooks that serve as examples or are part of our internal analyses. We plan to move them to a separate repository in the future. We also plan to create test data and a notebook showing how to run fine mapping. This part of the pipeline is indeed non-trivial.
However, if you are interested in reproducing the results of the L2G model, we suggest using already obtained fine-mapping results (credible set data), which are openly available to everyone here. This will save you a lot of time because fine-mapping is an extremely computationally demanding step.
Thank you for sharing the link @Yakov_Tsepilov! This will definitely save me a lot of time than re-running the whole fine-mapping pipeline.
L2G model uses functional data (such as DHS–promoter correlation) as features. I couldn’t find such data in the download section you shared. Is this data available somewhere? If not, does the pipeline include code to build this data?
Thanks.
Hi @Abdu_Omar , the latest L2G model does not use this data. The datasets used for the interaction features for the old model are outdated. We are currently working on including new interaction datasets in the model, which will be available in future releases.
Thanks for the update.
I’ll forward to the release of the latest datasets