schlosslab / sze_obesity_mbio_2016 Goto Github PK

@marcsze can you track down the papers that put human stool into mice and found changes in obesity and those that used antibiotics to do the same?

@marcsze can you redo the pull request to upload the figure as results/figures/flow_chart.png? The image you uploaded was of pretty poor quality and didn't replace the exiting Figure 1.

The supplemental figures may have gotten shuffled when adding the funnel plot

Hey Pat

I changed the strategy denovo R file and deleted the old one (although this shouldn't change anything on your end).

I also uploaded a newer version of the manuscript (with no meta at the end). I plan on editing this one for typos and such we have made.

Marc

First of all, thanks for having all the scripts here. It helps to understand your approach, and shows your commitment to open science.
I have a question on one of your steps. More specifically, I noticed in your scripts that you used a custom function for cross-validation instead of the one provided in the aucRF package. Could you elaborate on the reason? (You say there is a bug ). I am using it and I am surprised that the average cvAUC I get is way lower than I expected when I compared it to the one obtained from running my model with all the datapoints.

• all taxonomic ranks
• quantitative regression

@marcsze could you take a look at the PDF in submission and make sure everything looks good? Once I get word back from you, I'll update the line numbers in the Response to Reviewers document