Add get_slope_int_relationship_expr to get relationship between a pair of samples given slope and intercepts of lines to use as cutoffs.
#511
Conversation
…ilter` or `filtered_samples` are used
| .when( | ||
| kin_expr > duplicate_twin_min_kin, | ||
| hl.if_else( | ||
| dup_twin_ibd1_expr & (y_expr < parent_child_max_y), |
There was a problem hiding this comment.
So there are cases where the kinship coefficient is > 0.42 and (IBD0 or IBS0 or IBD0/IBS0) > some small number? I can't remember the figures but is this what triggered that very long investigation into what turned out to be oceania samples?
There was a problem hiding this comment.
Yeah, it's not really important for cuKING, but I added it for support with our PC-Relate IBD0, and yes I think it probably has to do with that long investigation
| min_related_hard_filter, | ||
| ) | ||
| gbi = relatedness_ht.annotate(s=list(relatedness_ht.key)) | ||
| gbi = relatedness_ht.annotate(s=[relatedness_ht.key[0], relatedness_ht.key[1]]) |
There was a problem hiding this comment.
Is this change for readability or does casting no longer work?
There was a problem hiding this comment.
list(relatedness_ht.key) returns ['i', 'j'] for each s, I'm assuming this was something that changed and was not always the case, but I couldn't confirm.
Fix
compute_related_samples_to_dropwhenmin_related_hard_filterorfiltered_samplesare used to build a list of additional samples to drop.