Research Data Management

• Basic Data Hygiene

• Metadata

Project Management

• File system organisation

• File naming

Link to handout: http://bit.ly/NHM_RDM_introduction

Yourselves!

be your own best friend:

aim to create secure materials that are easy to use and REUSE

Nine simple ways to make it easier to (re)use your data

We describe nine simple ways to make it easy to reuse the data that you share and also make it easier to work with it yourself. Our recommendations focus on making your data understandable, easy to analyze, and readily available to the wider community of scientists.

storify by Tom Webb @tomjwebb

Blog post by Dynamic ecology @DynamicEcology

Data carpentry

• Domain specific lessons available free online
  • Ecology materials
  • Genomics materials
  • Geospatial data materials
  • Biology semester long materials
• Look out for training sessions

Seek help from support teams

Most university libraries have assistants dedicated to Research Data Management:

@tomjwebb @ScientificData Talk to their librarian for data management strategies #datainfolit

— Yasmeen Shorish (@yasmeen_azadi) January 16, 2015

Plan your Research Data Management

• Start early. Make an RDM plan before collecting data.
  • RDM checklist
• Anticipate data products as part of your thesis outputs
• Think about what technologies to use

Data entering

excel: read only

@tomjwebb @tpoi excel is fine for data entry. Just save in plain text format like csv. Some additional tips: pic.twitter.com/8fUv9PyVjC

— Jaime Ashander (@jaimedash) January 16, 2015

@jaimedash just don’t let excel anywhere near dates or times. @tomjwebb @tpoi @larysar

— Dave Harris (@davidjayharris) January 16, 2015

Databases: more robust

• good qc and advisable for multiple contributors

@tomjwebb databases? @swcarpentry has a good course on SQLite

— Timothée Poisot (@tpoi) January 16, 2015

@tomjwebb @tpoi if the data are moderately complex, or involve multiple people, best to set up a database with well designed entry form 1/2

— Luca Borger (@lucaborger) January 16, 2015

Data formats

• .csv: comma separated values.
• .tsv: tab separated values.
• .txt: no formatting specified.

@tomjwebb It has to be interoperability/openness - can I read your data with whatever I use, without having to convert it?

— Paul Swaddle (@paul_swaddle) January 16, 2015

more unusual formats will need instructions on use.

Ensure data is machine readable

Use good null values

read.csv() utilities

• na.string: character vector of values to be coded missing and replaced with NA to argument eg
• strip.white: Logical. if TRUE strips leading and trailing white space from unquoted character fields
• blank.lines.skip: Logical: if TRUE blank lines in the input are ignored.
• fileEncoding: if you’re getting funny characters, you probably need to specify the correct encoding.

read.csv(file, na.strings = c("NA", "-999"), strip.white = TRUE, 
         blank.lines.skip = TRUE, fileEncoding = "mac")

readr::read_csv() utilities

• na: character vector of values to be coded missing and replaced with NA to argument eg
• trim_ws: Logical. if TRUE strips leading and trailing white space from unquoted character fields
• col_types: Allows for column data type specification. (see more)
• locale: controls things like the default time zone, encoding, decimal mark, big mark, and day/month names
• skip: Number of lines to skip before reading data.
• n_max: Maximum number of records to read.

read_csv(file, col_names = TRUE, col_types = NULL, locale = default_locale(), 
         na = c("", "NA", "-999"), trim_ws = TRUE, skip = 0, n_max = Inf)

Basic quality control

Raw data are sacrosanct

@tomjwebb don't, not even with a barge pole, not for one second, touch or otherwise edit the raw data files. Do any manipulations in script

— Gavin Simpson (@ucfagls) January 16, 2015

@tomjwebb @srsupp Keep one or a few good master data files (per data collection of interest), and code your formatting with good annotation.

— Desiree Narango (@DLNarango) January 16, 2015

Avoid catastrophe

Backup: the Open Science Framework osf.io

• version controlled
• easily shareable
• works with other apps (eg googledrive, github)
• work on an interface with R (OSFr) is in progress. See more here

Backup: Github

• most solid version control.

• keep everything in one project folder.

• Can be problematic with really large files.

Documenting your data

You got data. Is it enough?

@tomjwebb I see tons of spreadsheets that i don't understand anything (or the stduent), making it really hard to share.

— Erika Berenguer (@Erika_Berenguer) January 16, 2015

@tomjwebb @ScientificData “Document. Everything.” Data without documentation has no value.

— Sven Kochmann (@indianalytics) January 16, 2015

@tomjwebb Annotate, annotate, annotate!

— CanJFishAquaticSci (@cjfas) January 16, 2015

Document all the metadata (including protocols).@tomjwebb

— Ward Appeltans (@WrdAppltns) January 16, 2015

You download a zip file of #OpenData. Apart from your data file(s), what else should it contain?

— Leigh Dodds (@ldodds) February 6, 2017

metadata = data about data

Information that describes, explains, locates, or in some way makes it easier to find, access, and use a resource (in this case, data).

source: http://chiphouston.com/wp-content/uploads/2016/06/who-what-when-where-and-why11.jpg

Data Reuse Checklist

http://mozillascience.github.io/checklist/

Backbone of digital curation

Elements of metadata

• Structured data files:

  • readable by machines and humans, accessible through the web

• Controlled vocabularies eg. NERC Vocabulary server

  • allows for connectivity of data

Organising data and metadata

• Start at the very least by creating a metadata tab within your raw data spreadsheets

• Ideally set up a system of normalised tables (see section 3 in this post) and README documents to manage and document metadata.

• Ensure everything someone might need to understand your data is documented

• Different types data require different metadata

• When you’re ready to publish, structure metadata into an XML file, a searchable, shareable file.

Make your data alignable and generalisable

Bird Trait Networks dataset

• I’m using data from a project in which we compiled large dataset on bird reproductive, morphological, physiological, life history and ecological traits across as many bird species as possible to perform a network analysis on associations between trait pairs.

• I’ll use a simplified subset of the data to show a simple metadata (attribute) structure that can easily form the basis of a more formal EML (ecological XML) using function in the package EML

Data

• Like many real data sets, column headings are convenient for data entry and manipulation, but not particularly descriptive to a user not already familiar with the data.

• More importantly, they don’t let us know what units they are measured in (or in the case of categorical / factor data, what the factor abbreviations refer to). So let us take a moment to be more explicit:

Make an attribute table

I use functions in eml_utils.R to:

• create an attr_tbl in which to complete all info required
• to extract elements from attr_tbl to supply to EML generating functions.

library(RCurl)
eval(parse(text = getURL(
    "https://raw.githubusercontent.com/annakrystalli/ACCE_RDM/master/R/eml_utils.R", 
    ssl.verifypeer = FALSE)))

Create attr_tbl shell

• load data

dt   <- read.csv("data/bird_trait_db-v0.1.csv")

• create attr_tbl shell from your data (dt)

  • use function get_attr_shell from eml_utils.R.

attr_shell <- get_attr_shell(dt)

attr_tbl shell structure

str(attr_shell)

## 'data.frame':    6 obs. of  11 variables:
##  $ attributeName      : chr  "species" "max.altitude" "dev.mode" "courtship.feed.m" ...
##  $ attributeDefinition: logi  NA NA NA NA NA NA
##  $ columnClasses      : chr  "character" "numeric" "numeric" "numeric" ...
##  $ numberType         : logi  NA NA NA NA NA NA
##  $ unit               : logi  NA NA NA NA NA NA
##  $ minimum            : logi  NA NA NA NA NA NA
##  $ maximum            : logi  NA NA NA NA NA NA
##  $ formatString       : logi  NA NA NA NA NA NA
##  $ definition         : logi  NA NA NA NA NA NA
##  $ code               : logi  NA NA NA NA NA NA
##  $ levels             : logi  NA NA NA NA NA NA

attributes df columns

I use recognized column headers shown here to make it easier to create an EML object down the line. I focus on the core columns required but you can add additional ones for your own purposes.

Attributes associated with all variables:

• attributeName (required, free text field)
• attributeDefinition (required, free text field)
• columnClasses (required, "numeric", "character", "factor", "ordered", or "Date", case sensitive)

columnClasses dependant attributes

• For numeric (ratio or interval) data:
  • unit (required, see eml-unitTypeDefinitions and working with units)
• For character (textDomain) data:
  • definition (required)
• For dateTime data:
  • formatString (required) e.g for date 11-03-2001 formatString would be "DD-MM-YYYY"
    
• I use the columns code and levels to store information on factors. Use ";" to separate code and level descriptions. These can be extracted by eml_utils.R function get_attr_factors() later on.

Complete attr_tbl

write.csv(attr_shell, file = "data/attr_shell.csv")

attr_tbl <- read.csv(file = "data/attr_tbl.csv")

Completed attribute table attr_tbl

From raw to analytical data

Automated == reproducible

HOW?

Strategy against chaos

Use R projects

Keep your work tidy and self-contained

Raw data \(\rightarrow\) data

Pick a strategy, any strategy, just pick one!

data/
data-raw
data-clean

data/
  - raw/
  - clean/

Data \(\rightarrow\) results

Pick a strategy, any strategy, just pick one!

R

code

scripts

analysis

bin

  /Users/jenny/research/bohlmann/White_Pine_Weevil_DE:
  total used in directory 246648 available 131544558
  drwxr-xr-x  14 jenny  staff        476 Jun 23  2014 .
  drwxr-xr-x   4 jenny  staff        136 Jun 23  2014 ..
  -rw-r--r--@  1 jenny  staff      15364 Apr 23 10:19 .DS_Store
  -rw-r--r--   1 jenny  staff  126231190 Jun 23  2014 .RData
  -rw-r--r--   1 jenny  staff      19148 Jun 23  2014 .Rhistory
  drwxr-xr-x   3 jenny  staff        102 May 16  2014 .Rproj.user
  drwxr-xr-x  17 jenny  staff        578 Apr 29 10:20 .git
  -rw-r--r--   1 jenny  staff         50 May 30  2014 .gitignore
  -rw-r--r--   1 jenny  staff       1003 Jun 23  2014 README.md
  -rw-r--r--   1 jenny  staff        205 Jun  3  2014 White_Pine_Weevil_DE.Rproj
  drwxr-xr-x  20 jenny  staff        680 Apr 14 15:44 analysis/
  drwxr-xr-x   7 jenny  staff        238 Jun  3  2014 data/
  drwxr-xr-x  22 jenny  staff        748 Jun 23  2014 model-exposition/
  drwxr-xr-x   4 jenny  staff        136 Jun  3  2014 results/

Data

Ready to analyze data:

Raw data:

Analysis and figures

R scripts + the Markdown files:

sample_ready_to_analyze_data

The figures created in those R scripts and linked in those Markdown files:

sample_raw_data

Scripts

Linear progression of R scripts, and Makefile to run the entire analysis:

sample_scripts

Results

Tab-delimited files with one row per gene of parameter estimates, test statistics, etc.:

sample_results

Expository files

Files to help collaborators understand the model we fit: some markdown docs, a Keynote presentation, Keynote slides exported as PNGs for viewability on GitHub:

sample_expository

Caveats / problems with this example

• This project is nowhere near done, i.e. no manuscript or publication-ready figs

• File naming has inconsistencies due to three different people being involved

• Code and reports/figures all sit together because it’s just much easier that way w/ knitr & rmarkdown

Wins of this example

• Someone can walk away from the project and come back to it a year later and resume work fairly quickly

• Collaborators (the two other people, the post-doc whose project it is + the bioinformatician for that lab) were able to figure out what I did and decide which files they needed to look at, etc.

Evolution of your file system

• Be consistent – when developing a naming scheme for folders it is important that once you have decided on a method, you stick to it. If you can, try to agree on a naming scheme from the outset of your research project

• Structure folders hierarchically: - start with a limited number of folders for the broader topics - create more specific folders within these

• Separate ongoing and completed work: as you start to create lots of folders and files, it is a good idea to think about separating older documents from those you are currently working on

Prose

• Sometimes you need a place to park key emails, internal documentation and explanations, random Word and PowerPoint docs people send, etc.

• This is kind of like from_joe, where I don’t force myself to keep same standards with respect to file names and open formats.

Recap

File organization should reflect inputs vs outputs and the flow of information

/Users/jenny/research/bohlmann/White_Pine_Weevil_DE:
drwxr-xr-x  20 jenny  staff        680 Apr 14 15:44 analysis
drwxr-xr-x   7 jenny  staff        238 Jun  3  2014 data
drwxr-xr-x  22 jenny  staff        748 Jun 23  2014 model-exposition
drwxr-xr-x   4 jenny  staff        136 Jun  3  2014 results

Prepare data \(\rightarrow\) Do stats \(\rightarrow\) Make tables & figs

01_marshal-data.r
02_pre-dea-filtering.r
03_dea-with-limma-voom.r
04_explore-dea-results.r
90_limma-model-term-name-fiasco.r

02_pre-dea-filtering-preDE-filtering.png
03-dea-with-limma-voom-voom-plot.png
04_explore-dea-results-focus-term-adjusted-p-values1.png
04_explore-dea-results-focus-term-adjusted-p-values2.png
...
90_limma-model-term-name-fiasco-first-voom.png
90_limma-model-term-name-fiasco-second-voom.png

Names matter

What works, what doesn’t?

NO

myabstract.docx
Joe’s Filenames Use Spaces and Punctuation.xlsx
figure 1.png
fig 2.png
JW7d^(2sl@deletethisandyourcareerisoverWx2*.txt

YES

2014-06-08_abstract-for-sla.docx
joes-filenames-are-getting-better.xlsx
fig01_scatterplot-talk-length-vs-interest.png
fig02_histogram-talk-attendance.png
1986-01-28_raw-data-from-challenger-o-rings.txt

Three principles for (file) names

Awesome file names :)

Machine readable

Globbing

Excerpt of complete file listing:

Example of globbing to narrow file listing:

Same using Mac OS Finder search facilities

Same using regex in R

Include important metadata

e.g. I’m saving a number of files of extracted environmental data at different resolutions (res) and for a number of months (month).

write.csv(df, paste("variable_", res, month, sep ="_"))

df <- read.csv(paste("variable_", res, month, sep ="_"))

Recap: machine readable

Human readable

Example

Which set of file(name)s do you want at 3 a.m. before a deadline?

Embrace the slug

Recap: Human readable

Plays well with default ordering

Examples

Chronological order:

Logical order: Put something numeric first

Dates

Use the ISO 8601 standard for dates: YYYY-MM-DD

Left pad other numbers with zeros

If you don’t left pad, you get this:

10_final-figs-for-publication.R
1_data-cleaning.R
2_fit-model.R

which is just sad :(

Recap: Plays well with default ordering

• Put something numeric first

• Use the ISO 8601 standard for dates

• Left pad other numbers with zeros

Recap: Three principles for (file) names

1. Machine readable

2. Human readable

3. Plays well with default ordering

Go forth and use awesome file names :)

chronological_order

logical_order

Activity : Create an RStudio project

New Directory

File -> New Project -> New Directory

Empty Project

In the Project Type screen, click on Empty Project.

Name project directory

In the Create New Project screen, give your project a name and ensure that create a git repository is checked. Click on Create Project.

New project directory created

RStudio will create a new folder containing an empty project and set R’s working directory to within it.

Two files are created in the otherwise empty project:-

• .gitignore - Specifies files that should be ignored by the version control system.
• NHM_Advanced_comp_course.Rproj - Configuration information for the RStudio project

There is no need to worry about the contents of either of these for now.

Create a data folder

Success!

Acknowledgements