Ordering by Attribute seems to re-label / permute the samples incorrectly

Hello MEGAN team and community,

I am using MEGAN Community Edition 6.18.4 (January 29, 2020) on Mac OS 10.15.

I am comparing 77 individual rma6 files using the “compare” function. This is a case/control study, so per advice from a previous post on this community, I have created an attribute in the ‘sample viewer’ which codes 0 or 1 for ‘control’ or ‘case’. I next do two things. 1) ‘Color by attribute’ so that I have only two colors for case and control. 2) ‘Sort by attribute’ so that I can view all of the cases next to one another, and the controls next to one another in the ‘bar chart’ view.

I have noticed that this changes my underlying data.

Specifically, if I look at the bar chart of the read counts in a MEGAN file (at a specific node, say “Guillardia”) where I do none of the above attribute work, I see a certain pattern (i.e. sample “01.h38au.” has 77 reads to Guillardia, sample “02.h38au.” has 0 reads to Guillardia etc. I assume these are accurate.

After doing steps 1 and 2 above and viewing a bar chart, I see now that the quantities have been permuted across the 77 samples, e.g. sample “05.h38au” has 77 reads, etc. and sample “01.h38au” no longer has 77 reads.

I have attached a screenshot of the two bar charts.

The top is after doing steps 1 and 2, and then re sorting the rows so that they go in the order they were uploaded, using the function “sort rows” in sample viewer. The bottom is a new Megan session where the exact same data is read in (same order), and immediately the bar chart is produced. No attributes.

I have highlighted that the unique maximum read count (77) is associated with two different samples, depending on which plot you look at. What I expect, is that the plots should look exactly the same aside from different coloring, but they do not.

Thank you anyone who can provide some insight to this problem,


PS: A second example using another node

Yes, it appears that the ordering of the samples as changed.
Press on one of the buttons in the short second row of buttons to reorder the samples alphabetically or by counts.

It looks like sorting in the sample viewer doesn’t always carry over to the chart window and so you get discrepant orderings. I will look into this. For now, please use the work around described above.