Manhattan data interpretation

I’m trying to get the best gene prediction for each GWAS in the database. I downloaded parquet files from this directory: /pub/databases/opentargets/genetics/latest/manhattan and I’m trying to interpret the top10_genes_raw_ids and top10_genes_raw_score column. I’m wondering how is the raw score calculated and what should I do to pick for the best guess gene for each variant.

I would also like to confirm if the bestgene entry using api query contains the same information as the top10_genes. I also see a bestlocus2gene entry in the api query and I’m wondeirng if they are the genes with the highest L2G score.