Online Analysis

HARMONIES takes the subject-by-taxon count matrix and the subjects’ phenotype as the model input. Please refer to the example of microbiome count information on the right for how to organize the input information into a single .csv file.

HARMONIES can handle either one or two phenotypes among all the subjects. When all the subjects have the same phenotype, HARMONIES provides (1) the estimated partial correlations in the network, and (2) the normalized abundances given by the zero-inflated negative binomial model with the Dirichlet process prior. Given two phenotypes (e.g., subjects with disease and healthy controls), HARMONIES jointly models the observations from the two groups to achieve an improved abundance estimation. It provides (1) the estimated partial correlations in two networks corresponding to the two groups, (2) the normalized abundances of the taxa in each phenotype group, (3) the posterior probability for each taxon to be differentiating between the two groups, and (4) the estimated fold change of each taxon between the two groups.

Input your data

A valid email address is suggested to receive the notice when job is completed. (the average job execution is 1 - 10 minutes)
0 < input < 1
0 < input < 3
10 <= input <= 50
0.01 < input < 0.1
2 < input < 50
0.01 <= input <= 0.1
1 <= input <= 2e+08

Group name below should be less than 25 characters. (Only number and letters are allowed). Group name will display as filename of results.

e.g. group1
E.g. group2

Microbiome Count Information

A .csv file containing the observed microbiome counts and sample information:

  1. Each row corresponds to a sample in the study.
  2. The first two columns contain the sample ID and the group information.
  3. The 3rd to the last column consist the observed microbiome count matrix, with values being nonnegative integers.