hpgltools

Overview

A bunch of R functions to make playing with high throughput data easier, developed for applications at the UMD Host-Pathogen Genomics Laboratory (HPGL).

Although the functionality in this packaged was developed primarily to answer questions related to host-pathogen genomics and transcriptomics, much of the functionality is general enough that it may be useful for many other kinds of analyses.

Installation

There are too many ways to install software in R. The following are a few which I have performed for this package. This contains a mix of bioconductor and cran packages. The following instructions are what I used on a brand-new Debian unstable computer with nothing pre-installed.

Compiled Prerequisites

Quite a few packages used by hpgltools have compiled code, here is an incomplete list of requisites and the likely Debian-based.

• The basics: gcc,g++,gfortran,make,boost,build-essential,automake,autoconf
• curl
• libgvc
• libnlopt
• libxml2
• pandoc
• pkg-config
• texlive
• udunits2

## As the super user:

## For the basics:
apt-get install gcc g++ cc1plus gfortran gawk sed bison libboost-all-dev build-essential \
  curl libcurl4-gnutls-dev:amd64 libcurl4:amd64 automake autoconf autotools-dev \
  pkg-config

## For the prerequisite tools:
apt-get install udunits2-bin libudunit2-0 libudunits2-dev libnlopt0 libnlopt-dev \
  libxml2-dev pandoc texlive-latex-recommended texlive-science texlive-extra-utils \
  texlive-fonts-extra texlive-latex-extra

The actual R installation

Download the package via 'git pull' or from the github download link, go into the hpgltools/ directory and:

make prereq  ## install knitr, devtools, and friends.
make deps  ## Pick up dependencies from the DESCRIPTION file.
make install  ## Perform the installation
make check  ## Perform the R CMD check suite.
make vt  ## Rebuild the vignettes to make sure everything is ok.
make test  ## Run the tests

Other Makefile targets of interest

The provided Makefile has other targets which might be useful:

## Invoke R CMD build
make build
## Invoke R CMD check
make check
## Cleanup this tree
make clean
## Rerun roxygen, rebuild the vignettes, and the reference manual
make document
## Install this via R CMD INSTALL
make install
## Rebuild the vignettes
make vignette

Exploring

The easiest way to poke at this and see what it can do is:

library(hpgltools)
browseVignettes("hpgltools")

As of last count, there were a couple examples using the data(fission) set, pasilla, and a bacterial data set.

Functionality

Listed by function then filename in R/.

• annotations: Handles loading annotation data from sources like OrganismDb.
  • biomart: Uses the biomaRt interface to query ensembl.
  • genbank: Uses Rentrez to query genbank.
  • microbesonline: Uses some XML/MySQL to query microbesonline.org -- though I apparently queried them too often in testing from my workstation, so I got blacklisted :(
  • orgdb: Uses the DBI interface to query OrganismDbi/TxDb/OrgDb instances.
  • uniprot: Extract data from either uniprot text files or their webservices.
  • gff: Read gff files and take annotation data from them.
  • text: Extract putative annotation data from various text formats, primarily the csv files from trinity.
  • kegg: Use the KEGG webservices api.
• ExpressionSets: Attempts to simplify gathering together counts/annotations/designs.
  • tximport: Uses tximport when given filenames ending in supprted extensions.
  • read metadata: Able to read metadata from xlsx/csv/etc and extract count table filenames from them.
  • write expressionsets: Writes rdata files of the expressionsets along with xlsx files with the raw data, normalized data, and plots.
  • expt: A light S3 object wrapper around ExpressionSets to hopefully make them easier to play with.
• Helpers: Miscellaneous functions which don't fit elsewhere
  • installation: please_install() attempts to install packages from various sources, including biocLite(), install.packages(), install_github(), etc for given packages. Also provides a function to install all of bioconductor for the truly self-flagellating.
  • miscellaneous: Arbitrary (hopefully) helpful functions including sm() to silence most (all?) functions and pp() to fill in default values when making plots.
• Visualization and Sample Metrics: Plotters and distribution visualizers before doing expression analysis
  • bar: Some re-used bar plots like library size.
  • circos: Make using circos less obnoxious.
  • distribution: Plot some distributions/density plots.
  • dotplot: Plot some common dot plots.
  • genplot: Attempt to make genoplotR less annoying.
  • gvis: Small wrappers for gvis.
  • heatmap: Various heatmap shortcuts.
  • hist: Reused histograms.
  • misc: Some misc plots which have no other home.
  • point: Creates various reused scatter/point plots.
  • shared: Invokes a series of plots on new data.
• Model Tests / Batch Evaluation: Play with models, visualize the changes they evoke.
  • pca: Functions to simplify PCA analyses.
  • TSNE: Functions to simplify TSNE analyses.
  • surrogates: Different ways of evaluating surrogate variables(often batch) as per Leek et al.
  • model testing: Query the rank of various models.
  • varpartition: Use variancePartition to evaluate the contribution of experimental factors in data sets.
• Normalization: Various data normalizers under a single roof
  • batch: Batch correction via limma, sva, ruv.
  • convert: Perform cpm/rpkm/patternkm normalizations.
  • filter: Low-count filtering via cbcb and genefilter.
  • norm: Normalizations taken from cbcb, DESeq, and edgeR.
  • transform: Performs logn transformations of data.
  • shared: Calls functions in the norm_* files.
• Differential Expression: Functions to simplify invoking the various differential expression tools
  • basic: Completely naive differential expression, provides baseline for comparison.
  • deseq: Invokes DESeq2 for all pairwise comparisons of a data set.
  • edger: Invokes edgeR for all pairwise comparisons of a data set.
  • limma: Invokes limma for all pairwise comparisons of a data set.
  • shared: Shared functions across all tools.
• Ontology, KEGG, and friends: Query if given sets of genes are significantly over represented in various annotation schemes.
  • clusterprofiler: Invokes the clusterProfiler family of ontology searches.
  • goseq: Invokes goseq for ontology searches, helps graph results.
  • gostats: Invokes gostats for ontology searches, helps graph results.
  • gprofiler: Invokes gProfiler for ontology searches.
  • kegg: Functions to poke at KEGG enrichment.
  • reactome: Functions to poke at reactome enrichment.
  • topgo: Invokes topGO for ontology searches, graphs results.
  • shared: Calls functions in the ontology_* files.
• Random bits:
  • snp: Some functions to gather snp data into hopefully easier data structures.
  • proteomics: Parsers for mzXML data, plotters for the data they contain.
  • kmeans: A couple functions for kmeans clustering and visualization from Keith and Ginger.
  • motif: Some motif analysis wrappers.
  • nmer: Simplifying functions when working with nmers.
  • xlsx: Writing files for Excel is terrible, this hopefully makes it less so.