This patch: https://github.com/StatFunGen/pecotmr/pull/359/files is very useful reducing the memory footprint for the TWAS test pipeline but currently it is very rough estimate and counter intuitive --- users have to additionally calculate in their head the estimate for LD matrices size. We can improve this estimate to be more intuitive to guide users splitting by batch.