The BigQuery API does not currently support ingestion of file formats besides CSV (one can also ingest gzipped CSV files). We also currently don't support creation of tables with nested records.
In order to use data stored in protobufs, you would have to export your data to a CSV format. Any nested data would have to be "flattened."
However, the BigQuery team is interested in developer feedback regarding additional data ingestion formats. What is your use case? Are there other data formats you are interested in using?