PMCosm: Probabilistic Models for Microbial-Cosmos

This PMCosm package implements statistical inferences for microbial ecology analysis. This package provides methods for

categorizing the admixed structure of microbial communities,
detecting influential microbes in both engineered and natural environments,
evaluating sub-communities' preferences of life strategies,
identifying causal relationship between microbes' metabolism functions and fluctuations of surrounding environments,
predicting the community composition and structure for a given local environment.

Installation and documentation

To install, open R and type:

install.packages("devtools")
devtools::install_github("YushiFT/PMCosm")

Package overview

Functions

calc_mle_trio: Calculate MLE estimates for universal parameter trio for inferences about the admixed community structure.
classify_taxa: Traditional relative-abundance-based methods to classify microbial taxa into rare or abundant biospheres.
classify_vag_lag: Define a model-driven testable boundary between dispersal vangguards and dispersal laggards.
is_dispersion: Hypothesis test for the existence of dispersion.
is_overdispersion: Hypothesis test for the existence of over-dispersion.
is_zero_infla: Check the existence of inflated zeros.
pca_simple: A simplified principle component analysis for microbial data.
plot_trio: Graphic display of trio estimates to observe the Two-Wing pattern.

Quick start guide

Deriving MLE estimates for the admixed structure of environmental microbial communities

library(PMCosm)
# import sample data from microbial communities in hangzhou bay
data(hzmicrobe)
param_trio_bay <- calc_mle_trio(mic_bay, n_sample=10, replicates=3)
param_trio_era <- calc_mle_trio(mic_era, n_sample=12, replicates=3)

Visualize the Two-Wing admixed structure by trio parameters estimates

library(ggplot2)
library(latex2exp)
# for bay
plot_trio(param_trio_bay, point_size=0.8, a=0.6)
# zoom in for the majority 
plot_trio(param_trio_bay, point_size=0.8, a=0.6, zoom_in=TRUE)
# for era 
plot_trio(param_trio_era, point_size=0.8, a=0.6)

Categorize dispersal vanguards and laggards by identifying the model-driven testable boundary

id_vag_lag_bay <- classify_vag_lag(mic_bay, param_trio_bay)
id_vag_lag_era <- classify_vag_lag(mic_era, param_trio_era)
# print ids of dispersal vanguards in bay
print(id_vag_lag_bay$vanguards)
# print ids of dispersal laggards in bay
print(id_vag_lag_bay$laggards)
# print ids of dispersal vanguards in era
print(id_vag_lag_era$vanguards)
# print ids of dispersal laggards in era
print(id_vag_lag_era$laggards)

Visualize community structure as an admixture of dispersal vanguards and laggards

plot_vag_lag(param_trio_bay, id_vag_lag_bay)
plot_vag_lag(param_trio_era, id_vag_lag_era)