momr-package: Mining Metaomics Data In R

In eprifti/momr: Mining Metaomics Data (MetaOMineR)

Mining Metaomics Data In R

Description

momr also known as MetaOMineR is a R package that offers many functions and modules needed for the analyses of quantitative metagenomics data. It is conceived for the analyses of whole NGS data but can be used for 16S datasets as well or other type of omics data. Developed since the very beginning of the field the package has evolved and is structured around different modules such as preprocessing, analysis, visualization, etc. This package contains the different algorithms and routines as well as some test data objects. It is used along with other data packages that contain the needed information to describe a given catalogue developped in the same series.

Details

Package:	momr
Type:	Package
Version:	1.2
Date:	2016-09-14
License:	Artistic-2.0

The starting point of the analyses starts with a read count matrix that has been mapped onto a gene catalog. This raw read count matrix can be preprocessed through downsizing, normalization and filtering steps to obtain the abundance frequencies. The samples can then be clustered in different ways to check for similarity and outliers. Genes can than be statistically related to a given phenotype in order to select those that are of most interest (the biomarkers). Genes of interest can be projected onto the MGS catalog to obtain a reduced dataset of microbial entities that is to be further annotated.

Updates

2016/09/14:	Updating package with a number of fixes and upgrades
2015/10/10:	Fix bug in the downsize.matrix function and parallel computing
2015/06/24:	First version for CRAN. One year long changes implemented, map reduce, normalization etc...
2014/03/24:	First official release, licence added, and map-reduce procedures
2014/02/03:	added new functions phenoPairwiseRelations, extractSignificant and lmp
2014/02/03:	testRelations modified to give the direction of a correlation
2014/02/03:	MGS sample catalog udate. This version has genes sorted based on the whole metahit cohort

Author(s)

Authors:	Edi Prifti and Emmanuelle Le Chatelier
Maintainer:	Edi Prifti <edi.prifti [at] gmail.com>

References

Le Chatelier, Emmanuelle, Trine Nielsen, Junjie Qin, Edi Prifti, Falk Hildebrand, Gwen Falony, Mathieu Almeida, et al. "Richness of Human Gut Microbiome Correlates with Metabolic Markers." Nature 500, no. 7464: 541???546.

Examples

  
# load the package
library(momr)

#' all the data in the package
# data(package="momr")

#' load the raw and frequency test dataset
data("hs_3.3_metahit_sample_dat_raw")
data("hs_3.3_metahit_sample_dat_freq")


#' NORMALIZATION
#' This should be performed with the whole dataset (complete catalogue). 
#' But here is an exemple with the subset of the data for illustration purposes
data(hs_3.3_metahit_genesize)
norm.data <- normFreqRPKM(dat=hs_3.3_metahit_sample_dat_raw, cat=hs_3.3_metahit_genesize)


#' CLUSTERING OF SAMPLES
hc.data <- hierClust(data=hs_3.3_metahit_sample_dat_freq[,1:5], side="col")
clust.order <- hc.data$mat.hclust$order
#' order samples followin the hierarchical clustering
ordered.samples <- colnames(hs_3.3_metahit_sample_dat_freq[,1:5])[clust.order]
#' how close are the two first samples (spearman, rho)
hc.data$mat.rho[ordered.samples[1], ordered.samples[2]]
# select the samples closely related together
close.samples <- filt.hierClust(hc.data$mat.rho, hclust.method = "ward", plot = TRUE, filt = 0.5)

#' CLUSTER GENES ON THE MGS CATALOG
#' load the curated mgs data for the hs_3.3_metahit catalog
data("mgs_hs_3.3_metahit_sup500")

#' project a list of genes onto the mgs
genebag <- rownames(hs_3.3_metahit_sample_dat_freq)
mgs <- projectOntoMGS(genebag=genebag, list.mgs=mgs_hs_3.3_metahit_sup500)

#' extract the profile of a list of genes from the whole dataset
mgs.dat <- extractProfiles(mgs, hs_3.3_metahit_sample_dat_freq, silent=FALSE)

#' plot the barcodes
par(mfrow=c(5,1), mar=c(1,0,0,0))
for(i in 1:5){ 
  plotBarcode(mgs.dat[[i]])
}

#' compute the filtered vectors
mgs.mean.vect <- computeFilteredVectors(profile=mgs.dat, type="mean")


#' TEST RELATIONS
#' for the first 1000 genes
res.test <- testRelations(data=hs_3.3_metahit_sample_dat_freq[1:500,],
                          trait=c(rep(1,150),rep(2,142)),type="wilcoxon")
head(res.test)
print(paste("There are",sum(res.test$p<0.05, na.rm=TRUE),"significant genes and",
            sum(res.test$q<0.05, na.rm=TRUE), "after adjustment for multiple testing"))

res.test.mgs <- testRelations(data=mgs.mean.vect,trait=c(rep(1,150),rep(2,142)),type="wilcoxon")


#' DOWNSIZING
#' downsize the matrix
data.downsized <- downsizeMatrix(data=hs_3.3_metahit_sample_dat_raw[,1:5], repetitions=1, level=600)
colSums(data.downsized, na.rm=TRUE)

#' downsize the genecount
data.genenb <- downsizeGC(data=hs_3.3_metahit_sample_dat_raw[,1:5], level=600, repetitions=3)
par(mfrow=c(1,1), mar=c(4,4,4,4))
plot(density(colMeans(data.genenb, na.rm=TRUE)), main="density of downsized gene richness")

#' End of test file

eprifti/momr documentation built on Sept. 27, 2022, 3:36 a.m.