eclarke

plot_ggheatmap <- function(obj, n=nrow(obj), norm=TRUE, log=TRUE, 
                           colnames.in.pdata="NewSampleID", 
                           col.labels=NULL,
                           row.labels=NULL,
                           facet.by=NULL,
                           annotate.cols=NULL,
                           dendrogram=FALSE,
                           col.annotation.offset=1,
                           col.annotation.width=4) {
  ##

eclarke

#!/bin/bash

curl --header 'Access-Token: <your_access_token_here>' \
     --header 'Content-Type: application/json' \
     --data-binary '{"body":":)","title":"$@","type":"note"}' \
     --request POST \
     https://api.pushbullet.com/v2/pushes

eclarke

(defun ensure-in-vc-or-checkin ()
  (interactive)
  (if (file-exists-p (format "%s" (buffer-file-name)))
      (progn (vc-next-action nil) (message "Committed"))
    (ding) (message "File not checked in.")))

(defun export-bibtex ()
  "Exports Papers library using a custom applescript."
  (interactive)
  (message "Exporting papers library...")

eclarke

Keybase proof

I hereby claim:

• I am eclarke on github.
• I am erikclarke (https://keybase.io/erikclarke) on keybase.
• I have a public key ASDtTlFKxCJoOWfwfhHy-MJtbmGKnvzo-zxrJETJmi1kNQo

To claim this, I am signing this object:

eclarke

prop_presence_absence <- function(otu.pa, groups) {
  # Creates a proportional presence-absence melted dataframe suitable for use in
  # ggplot heatmaps to show varying within-group proportions of species.
  # 
  # Args:
  #   otu.pa: Matrix of presence-absence data. Columns are samples, rows are 
  #           species.
  #   groups: Grouping data frame. A column named "SampleID" should be unique 
  #           list of sample identifiers that match the column names of otu.pa. 
  #           The other column, named "group", should correspond to the group 

eclarke

# This is the testing function that builds the models
  testMixedEffects <- function(test.col) {
    # We return this default in case of error/no fit
    default <- list(null.model=NA, model=NA)
    
    counts <- .mat[, colnames(.mat)==test.col]
    if (sum(counts > 0)/length(counts) < sparsity) {
      message(sprintf("%s: too sparse, skipping", test.col))
      return(default) 
    }

eclarke

# Using indicspecies with a melted data frame

# Their input is actually not hard to work with. First we need to re-create the count matrix.
# This creates a data frame with the sample ID and study group as the first two columns, then each column after that is an OTU name
# (Replace column names as appropriate)
mat <- melted.df %>% reshape2::dcast(SampleID + StudyGroup ~ otu)

# Next, we actually convert things into inputs
# Hadley's stuff hates rownames, so we have to remake them
rownames(mat) = mat$SampleID

eclarke

docs:
	Rscript -e "devtools::document(roclets=c('rd', 'collate', 'namespace', 'vignette'))"

gh-pages:
	git checkout gh-pages
	git merge master -X theirs -m "merge master"

site:docs gh-pages
	Rscript -e "staticdocs::build_site(site_path='.', launch=FALSE)"
	git commit -am 'updated docs'

eclarke

#!/usr/bin/env python
"""Usage: python taxonomy_fixer.py [FILE]

Converts an ITS taxonomy file to eliminate taxa marked as unidentified,
swaps [kpcofg]__unidentified;s__Fungi to k__Fungi, and eliminates species
taxa that are simply s__[genus]_sp.

Writes to stdout.
"""
# Erik Clarke <ecl@mail.med.upenn.edu>

eclarke

# Indicator value functions -----------------------------------------------

#' Returns the indicator value (Dufrene, 1997) for a given row of species counts
#' along with a vector of class assignments.
#' @param row:     vector of counts (usually a row in a counts matrix)
#' @param class:   which level in the grouping variable to test
#' @param classes: factor describing the grouping of the counts vector
.indval <- function(row, class, classes) {
  idxs <- classes == class
  A.ij <- sum(row[idxs]) / sum(row)