Fig 5

[antagomir]

• Caption should cite the utilized data sets.
• Font sizes and should be larger for figures and subfigure letters (a, b, c..)
• Can we add circular loading plot to Fig 5a?

[antagomir]

@0xMuluh and @RiboRings could one of you help finalize this one?

[TuomasBorman]

@raivo-otus Can you add the circular loading plot? Use RDA results.

[raivo-otus]

I will look into it.

Just to confirm;
is the intent to replace the phylogenetic relationship graph with the circular loadings or add it alongside somehow?

[TuomasBorman]

You could replace it as the loading plot has the tree in the middle

[raivo-otus]

Facing difficulties getting the circular loading plot drawn.

Using the TSE and RDA results as they are, results in an error: "Too many features to plot."
Calculating RDA for a higher level—say altExp(tse, "Phylum")—and trying to plot results in an error with tree.

Error: <ggtree> object properties are invalid:
- @mapping must be <ggplot2::mapping>, not S3<data.frame>

Am I missing something crucial?

# In .rds creation chunk... 
altExp(tse, "Phylum") <- addRDA(
    altExp(tse, "Phylum"), assay.type = "clr", method = "euclidean",
    formula = x ~ Age_Months + Country, na.action = na.exclude
    )

# In plotting chunk... 
p_load <- plotLoadings(
  altExp(tse, "Phylum"),
  dimred = "RDA",
  layout = "heatmap",
  add.tree = TRUE
)

[TuomasBorman]

Facing difficulties getting the circular loading plot drawn.

Using the TSE and RDA results as they are, results in an error: "Too many features to plot." Calculating RDA for a higher level—say altExp(tse, "Phylum")—and trying to plot results in an error with tree.

Error: <ggtree> object properties are invalid:
- @mapping must be <ggplot2::mapping>, not S3<data.frame>

Am I missing something crucial?

# In .rds creation chunk... 
altExp(tse, "Phylum") <- addRDA(
    altExp(tse, "Phylum"), assay.type = "clr", method = "euclidean",
    formula = x ~ Age_Months + Country, na.action = na.exclude
    )

# In plotting chunk... 
p_load <- plotLoadings(
  altExp(tse, "Phylum"),
  dimred = "RDA",
  layout = "heatmap",
  add.tree = TRUE
)

This might be related to ggplot and ggtree updates. Can you install ggtree from GitHub? They updated it just lately.

[raivo-otus]

Facing difficulties getting the circular loading plot drawn.
Using the TSE and RDA results as they are, results in an error: "Too many features to plot." Calculating RDA for a higher level—say altExp(tse, "Phylum")—and trying to plot results in an error with tree.

Error: <ggtree> object properties are invalid:
- @mapping must be <ggplot2::mapping>, not S3<data.frame>

Am I missing something crucial?

# In .rds creation chunk... 
altExp(tse, "Phylum") <- addRDA(
    altExp(tse, "Phylum"), assay.type = "clr", method = "euclidean",
    formula = x ~ Age_Months + Country, na.action = na.exclude
    )

# In plotting chunk... 
p_load <- plotLoadings(
  altExp(tse, "Phylum"),
  dimred = "RDA",
  layout = "heatmap",
  add.tree = TRUE
)

This might be related to ggplot and ggtree updates. Can you install ggtree from GitHub? They updated it just lately.

Updating packages fixed the issue.

However the circular plot with loadings, is not very legible. Not sure how to fiddle with the parameters to improve clarity. I feel that the space in the collection of plots (fig5-A) feels a bit cramped for this style of plot.

[TuomasBorman]

How about 2 components and top 20 taxa?

Also, I the coloring of taxa is not working. It is probably a bug --> can you check why and report to miaViz

[raivo-otus]

ncomponents and n options do nothing. and throw a warning

1: Unknown or uninitialised column: `subgroup`. 

[raivo-otus]

How about 2 components and top 20 taxa?

Also, I the coloring of taxa is not working. It is probably a bug --> can you check why and report to miaViz

Okay. Will attempt to investigate and get an issue up.

[TuomasBorman]

The version in the branch should already work (at least to some extent). You could install the miaViz from the branch and modify the manuscript. Also can you post the figure here so we can check if it is better than the current one.

We can later optimize the miaViz (there is no big hurry on that, and I think it would be good task for you)