plot_region_expr <- function(chrom, start.pos, end.pos, biomart, salmon, sequenza_df, ...) {
  genes <- 
    getBM(
      attributes = c("chromosome_name", "start_position", "end_position", "hgnc_symbol"),
      filters = c("chromosome_name", "start", "end"),
      values = list(chromosome_name = sub("^chr", "", chrom), start = start.pos, end = end.pos),
      mart = biomart) %>% 
    subset(.$hgnc_symbol %in% rownames(salmon$clean$counts)) %>% 
    arrange(start_position) %$%
    hgnc_symbol
  
  annot <- 
    sequenza_df %>% 
    filter(chromosome == chrom, end > as.numeric(start.pos), start < as.numeric(end.pos)) %>% 
    group_by(sample) %>% 
    summarise(segmean = mean(segmean) + 2) %>% 
    mutate(patient = get_patient_id(sample)) %>% 
    select(patient, CN = segmean) %>% 
    as.data.frame() %>% 
    column_to_rownames("patient")
  
  cbs <- setdiff(colnames(salmon$clean$cvst), rownames(sequenza_df))
  
  annot <- rbind(annot, data.frame(CN = rep(2, length(cbs)), row.names = cbs))
  
  patient_order <- annot %>% 
    rownames_to_column("patient") %>% 
    arrange(CN) %$% 
    patient %>% 
    intersect(colnames(salmon$clean$counts))
  
  pheatmap::pheatmap(t(assay(salmon$clean$cvst)[genes, patient_order]),
                     cluster_cols = FALSE, cluster_rows = FALSE, annotation_row = annot, ...)
  
  amp_patients_idx <- annot$CN > 2.1
  
  map(genes, ~assay(salmon$clean$cvst)[.x,]) %>% 
    set_names(genes) %>% 
    map(~wilcox.test(.x[amp_patients_idx], .x[!amp_patients_idx], alternative = "greater")) %>% 
    map_dbl("p.value")
}

pvals <- plot_region_expr("chr1", "156342801", "157204569", biomart, salmon,
                          sequenza_1mb_df, scale = "column")