3 ms·
It goes far beyond "chr1" = "1". There are many other fields with next-to-impossible to interpret data. The consortia have, at best, advisory roles, they cann
by dekhn 10y ago
It goes far beyond "chr1" = "1". There are many other fields with next-to-impossible to interpret data. The consortia have, at best, advisory roles, they cannot enforce any rules. And given an extensible file format, third parties are going to extend it with poorly defined/standardized data. Then the consumers have to harmonize the data. Data harmonization (or cleaning) it typically one of the bottleneck steps in applying machine learning at scale.