Total read counts or proportion

[wangjiawen2013]

Dear,
Abundance of taxon was measured with total reads counts in Sunbeam Final Report's heatmap while it was proportion in the original sunbeam paper (Clarke EL, 2019, Figure 3d). It makes me confused, So which one is better ?

[wangjiawen2013]

And, I tried to run sbx_report, the results looked wired:

2.1 Phylum-level heatmaps
Showing taxa at the phyla level or higher that appear in at least 1 samples.

No phyla present in requisite number of samples

2.2 Family-level heatmap
Showing taxa at the family level or higher that appear in at least 1 samples.

No phyla present in requisite number of samples

2.3 Genus-level heatmap
Showing taxa at the genus level or higher that appear in at least 1 samples.

No genera present in requisite number of samples

2.4 Species-level heatmap
Showing all species level annotations for all samples. Warning: this plot can be quite large.

Note that read counts was float type !
But I could see the taxon abundance in all_samples.tsv, so could you give me some suggestions on what caused the wired report ? Was it caused by the failure of the DNA library ?

[louiejtaylor]

Hi @wangjiawen2013--regarding your first question, you can use either metric. Relative abundance is good to express your data as parts of a whole, but does not provide information about absolute abundance. The one to use depends on your dataset and research question.

Regarding your output--that does look a bit odd. How many samples are you running? My hunch is that we look for taxa present in >1 samples, and you've only run one sample? We do count absolute abundance (number of reads) here, it's just the legend that's showing fractional readcounts.