Skip to content

v1.1.3: LRT and updated mash ordering
Compare
Choose a tag to compare
@johnlees johnlees released this 05 Apr 08:54
v1.1.3
0799574

This release adds the following features and fixes

  • Adds the likelihood ratio test (LRT) as the default p-value for binary traits (issue #42). This has higher power when effect sizes are large.
  • Ensure consistency of sample order when using mash rather than kmds. Previous releases will have assumed mash ordering was the same as seer; the .samples file is written correctly by the perl script and mitigates this.
  • Fix for the t-test in continuous traits, which could give incorrect p-values for negative effect sizes (issue #46).
  • Check for header with filter_seer and map_back

If using mash rather than kmds, please use this release or later.

Assets 5
Loading