@constantAmateur I think your last analysis demonstrates the underlying phenomenon well: by dropping a strict subset of genes (even just a few) you are forcing those cells to be comparatively highly weighted nearest neighbors in the KNN graph which the UMAP builds on (Iโd wager)
@johnny_israeli@rapidsai @lal_avantika @cjnolet The performance look amazing; these kinds of speed-ups will have a huge impact on how we do some of our analyses...awesome stuff!
@Jack_Kosmicki @saralpulit @ceclindgren@allison_halff Same! Iโve just restarted a breath of the wild replay and am falling in love all over again. @konradjk texting me his progress updates made me jealous! I also really liked the switch fire emblem game if yโall are taking recs
@A_A_Zaidi @CharlestonCWKC@SaharMozaffari In the last few years weโve somewhat adopted the terms eSNP / eGene to refer to the separate components, which might lead you to assume eQTL meant both? Thatโs the best Iโve got
@aaronquinlan My undergrad genetics professor only used chalkboards and on rare occasion an overhead projector for laminated notes heโd then further markup (sheets had punnet square & pedigree templates); infinitely superior to slides, although more work for us