(PM Session) Foundational Open Science Skills (FOSS) Lesson 3: Intro to Data Management

Date: 2020-09-28
Today Lead Instructor: Michele
Today Helpers: Tina, Tyson
Course Website: https://cyverse-learning-materials.github.io/foss
Zoom Link: https://arizona.zoom.us/j/86152278453

Instant Feedback: (please complete before you leave class) Complete Form

Agenda

Warm-up (5 minutes):

Questions & Comments about Project Management left over last week?

Introduction to Data Management (50 minutes)

Self Assessment

If you give your data to a colleague who has not been involved with your project, would they be able to make sense of it? Would they be able to use it properly?

• omg probably not 100% but i did try to make it easy to understand for other people but there'll be times where you need to present something, deadlines come up, and you are not doing your best documenting. I always tried to go back after deadlines and re-write stuff, but it's october and im tired lol.
• Nope. My boss frequently says she isn't allowed to touch any of our data because she might mess it up. I would hope it's intuitive, but I've never actualy tried!
• Definitely not 100%. That get's demonstrated when new scientists enter the lab and need to pick up where projects left off.
• Some projects yes, most may be not. There is room for improvement for sure, but also writing to make it understandable for all skill levels is a challenge.
• Tahsin: At this stage, No! But when I share my data to my colleague, I make sure it is understandable.
• I would give it a 50/50 shot. I try to have organization in my folders and background info docs. As well as some code comment
• Not at the moment. I am mostly in the exploratory phase of a new project so I am looking into a lot of new datasets and haven't spent the time to organize it all yet because I am not sure what I will end up using.
• No, because they don't use coding languages. If they did, I think my code would be interperable
• Some of my data is has metadata with it or it is downloaded from a government facility where one can find the metadata. But for data that I maniputlate it can be hard to understand for other people or future me.
• If they were preemptively familiar with the work tools that I use they could but other wise it could be a struggle. They would not know what I planned to use the data for so most likely they would not be able to use it properly.
• It depends on the dataset/project
• I suspect that they would be able to understand the functions that my code are performing (I am somewhat religious about commenting my scripts) on the data however understanding the data itself requires some background theory. The code and the dataset are not developed enough yet to be a standalone/self-contained entity from which anyone can understand the work.

If you come back to your own data in five years, will you be able to make sense of it? Will you be able to use it properly?

• I believe so… but that's such a long time
• Most likely I could remind myself what I was doing and use past documents I have saved in the cloud, git, and google to figure out what I was doing and what I wanted to do.
• I think that I am improving in this area, but not all of my past work is very well organized.
• I think so, I've gotten better at commenting out code and explaining what is happening/where to find what in different folders.
• Some datasets probably, but code not always…sometimes hard to find the code that I know I wrote (aka organization is not great)
• I dont think so. Right now I can't even remember most of my data after 6 months!
• I think so
• No, I can't do this even within the few weeks of peeer review, usually.
• yes. for sure.
• for the most part yes, I have gone back and had trouble but try my best to write detailed notes or versions of data/code/files
• I hope so, I think so after a day or refamiliarize with it.
• Perhaps after a few days of file exploration, reading past notes, and toying around with the code.

When you are ready to publish a paper, is it easy to find all the correct versions of all the data you used and present them in a comprehensible manner?

• yes. at least I got this one right lol
• yes, with a little bit more organization as I finish things up
• Yes
• No, I am awful and organinizing my data. Largely because the data itself is simulated and I am still working on getting the code in order to properly generate a good data set.
  when I'm ready to publish yes, but not always along the way :(
• Yes! I keep the original data in it's own folder, and the final versions with the final manuscript in it's own folder.
• yes
• No, I always have this moment like oops all the results were a little off when I finish a project because I go through and try to find all the right datasets and run again :/
• I have many version controls,
• Yes, I label all the things. Details, details, details. But it is a headache to keep track if I do not modify/notate in the moment.
• Yes, if I am publishing a paper I would make sure to have all my data available in a comprehensive manner because it frustrates me when I can't reproduce papers I read
• I am getting better at this!
• Yes, I tend to clean up my data right before publication, but I'm a mess leading up to that point

---

Breakout session 1 (Rms 1 and 2)

1. What are the two or three data types that you most frequently work with?

• Think about the sources (observational, experimental, simulated, compiled/derived)
• Also consider the formats (tabular, sequence, database, image, etc.)

• sound recordings (will be transcribed to text), texts (will be coded to csv), survey responses (csv)
• Large, public datasets
• Experimental data (stored in CSV files)
• Simulated data for computational experiments - for models under active development, so constantly rewriting code and re-gnerating data
• Observational, survey, clinical, health datasets (patient-reported outcomes)
• Weather data
• Biological samples (e.g., hormone measures), physiological measurements (e.g., cardiac data)
• Genomic data
• Time-series data (arrival statistics, video)
• Qualitative interview data that is transcribed and coded
• genomic (experimental) data

2. What is the scale of your data?

• file size (kb,mb,gb,tb,pb,eb), file #s, processing time, etc

• Observations in the hundreds or thousands (100-10,000 range, MB)
• sound recordings less than 100, texts (?? how to quantify this), survey responses less than 5000
• GBs to TBs
• ~58 terabytes, ~14 million images
  
• Small code files, <5 Gb simulated data files (csvs etc)

Best practices

• Project Charter (Planning)
• Consent (informed)
• standardized variables
• Metadata fields (I think there are some NSF standards)
• Redundancy
• Analysis metrics that are the industry norms/standards (Analyze)
• Persistence
• Instrument calibration
• Explicit units description
• for preserve -> backups of raw data protected with read only permissions
  ** (ahh this is a good idea)
• describe -> creating data dictionaries
  ** I like that idea ^

---

BioBreak (5 minutes)

Data Management Cont.

Discussion on Data Strategies

What is your strategy for storing and backing up your data?

dropbox
boxHealth
github
cyverse
Box
external drive

• well Greg just told me to do hard drive x2 (plus cloud and 2 physical copies) so I'll be doing that now lol
• UA Box , REDCAP, Google Drive
• supplementary materials section as part of journal publication submission
• HPC, lab server

how do you preserve your data?

Github (same)
Zenodo
HMP DACC
It goes into a local drive (becomes dark data) to die… <- who's this

Have you ever used or created a Data Management Plan for your research?

• what was the DMP used for?

• nope, but I want to start one for my second-year paper
• never done a formal DMP before but am currently working on one

FAIR

• FAIR is great, it's a good scaffold to work with
• FAIR seems more to do with technology, and CARE seems more to do with people. Both are needed!

Introduction to CyVerse

https://user.cyverse.org - User Portal (Account Creation)

https://learning.cyverse.org/what_is_cyverse/

Homework Assignment

Work through Self-Paced CyVerse Intro: https://cyverse-learning-materials.github.io/cyverse_mooc/

Links shared in class

Ontology v. Schema: https://www.w3.org/wiki/SchemaVsOntology
ORCID: https://orcid.org/
How to FAIR: https://howtofair.dk/how-to-fair/metadata/
Checksum: https://en.wikipedia.org/wiki/Checksum
FAIR sharing: https://fairsharing.org/
Creative Commons (licensing): https://creativecommons.org/
Licensing FAIR data for reuse: https://direct.mit.edu/dint/article/2/1-2/199/10013/Licensing-FAIR-Data-for-Reuse
FAIR Data assessment tool: https://ardc.edu.au/resource/fair-data-self-assessment-tool/
DMP Tool: https://dmptool.org/
Data Stewardship Wizard: https://ds-wizard.org/