How can I rename the columns in parquet to be consumed by BigQuery?

Question

We have 10K parquet files, resulting from a week of processing in a Google cluster. The resulting columns (historically) started with a number, like this: 2_xxxx