The database views marker_positive_cell_count_by_study_specimen and its associates are expensive to create/refresh (often crashing pg_restore and similar db management operations) and don't offer much functionality. They are only used to support the precomputed t-tests between predefined phenotype fractions. But these fractions can be computed much more quickly now that we have the dedicated counts service.
For this issue, deprecate marker_positive_cell_count_by_study_specimen and related views, replacing fraction_stats with a table that is created at initialization using the more optimized counts service functions.
The database views
marker_positive_cell_count_by_study_specimen
and its associates are expensive to create/refresh (often crashing pg_restore and similar db management operations) and don't offer much functionality. They are only used to support the precomputed t-tests between predefined phenotype fractions. But these fractions can be computed much more quickly now that we have the dedicated counts service.For this issue, deprecate
marker_positive_cell_count_by_study_specimen
and related views, replacingfraction_stats
with a table that is created at initialization using the more optimized counts service functions.