Fix cluster ordering in scanpy output

code/run-scanpy.R

@@ -95,6 +95,12 @@ run_scanpy <- function(sce, pars) {
   result <- dplyr::bind_cols(cluster, !!!out)
   result <- dplyr::mutate(result, raw_statistic = 0)
 
+  # In the past the output from Scanpy grouped the results for each cluster
+  # together with genes ranked. At some point however this changed for an
+  # unknown reason, so we rectify this here. We DO NOT change the ranking of
+  # scanpy.
+  result <- dplyr::arrange(result, cluster)
+
   # FIXME: Can we save the raw result in the anndata object?
   list(time = time, result = result, raw_result = result, pars = pars)
 }