@zacharylipton In single-cell biology there were a bunch of methods using zero-inflated likelihoods for counts data to deal with excess zeros (missing values due to technical artifacts) but in the end they may not be that useful over negative binomials htt
2,009 followers
21,631 followers
JAX's Churchill lab (@kwangbom @genomequant) applied a Bayesian model selection approach to unambiguously demonstrate 0️⃣ inflation in multiple biologically realistic scRNA-Seq datasets. Out now in @GenomeBiology: https://t.co/z3VZBBD4Ih https://t.co/EV06
3,301 followers
RT @jacksonlab: JAX's Churchill lab (@kwangbom @genomequant) applied a Bayesian model selection approach to unambiguously demonstrate 0️⃣ i…
141 followers
"A UMI count of zero does not necessarily imply that the gene is not expressed. This has led some...to propose imputation methods that convert zeros to non-zero values before analysis...we and others recommend against...replacing zeros in data with imputed