Correlation Heatmap for top 20 taxa (Genus/Species)

[msh-04]

Hi David,
Thanks for developing a handy package, "microViz". I'm not a proficient R expert and have struggled to get a Spearman correlation heatmap for the top 20 taxa (Genus/Species). I was able to create a correlation heatmap for "order level" using the code below; however, when I changed to the Genus level, it was so messy because of too many genera. Can I select the top 20 genera or species for this correlation heat map? If so, please let me know what code I can use in this case. I would greatly appreciate your help. Thanks.

ps_microviz <- ps %>%
ps_mutate(
C = if_else(Group == "Center", true = 1L, false = 0L),
S = if_else(Group == "South", true = 1L, false = 0L),
N = if_else(Group == "North", true = 1L, false = 0L),
E = if_else(Group == "East", true = 1L, false = 0L),
W = if_else(Group == "West", true = 1L, false = 0L)
)
taxa_cor_heatmap <- ps_microviz%>%
tax_agg("Order") %>%
tax_transform(trans = "hellinger", rank = "Order") %>%
tax_filter(min_prevalence = 5) %>%
cor_heatmap(
colors = heat_palette(palette = "Blue-Red 3"),
vars = c("C", "S", "N", "E", "W"),
cor = "spearman"
)

[msh-04]

Hi David,
Based on one of your examples for the "dietswap" data, I managed to get the correlation heatmap for the top genus level, as you can see in the attachments. I used the code below for this correlation heatmap. But I am confused about a variable (year) that I did not mutate in the coding. Could you suggest how I could get rid of this variable from the heatmap? Thanks for your help.

Cor_Heatmap_Genus_Pearson.pdf

psq <- ps %>%
ps_mutate(
Center = if_else(Group == "Center", true = 1L, false = 0L),
South = if_else(Group == "South", true = 1L, false = 0L),
North = if_else(Group == "North", true = 1L, false = 0L),
East = if_else(Group == "East", true = 1L, false = 0L),
West = if_else(Group == "West", true = 1L, false = 0L)
) %>%
tax_fix(unknowns = c("NA")) %>%
tax_filter(
tax_level = "Genus", min_prevalence = 1 / 10, min_sample_abundance = 1 / 10
) %>%
tax_transform("identity", rank = "Genus")

Proportional min_prevalence given: 0.1 --> min 6/60 samples.

randomly select 30 taxa from the 50 most abundant taxa (just for an example)

set.seed(123)
taxa <- sample(tax_top(psq, n = 50), size = 30)

draw the heatmap

cor_heatmap(
data = psq, taxa = taxa,
taxon_renamer = function(x) stringr::str_remove(x, " [ae]t rel."),
tax_anno = taxAnnotation(
Prev. = anno_tax_prev(undetected = 50),
Log2 = anno_tax_box(undetected = 50, trans = "log2", zero_replace = 1)
)
)