Biometadata Crashcourse

First steps towards FAIR (meta)data descriptions, ontology terms and data deposition

Maja Magel

Charlie Pauvert

2024-12-04

Thank you for joining!

Biometadata Crashcourse: First steps towards FAIR (meta)data descriptions, ontology terms and data deposition

About us

Lectures created by

• Charlie Pauvert
• Maja Magel
• input from helpers & participants of previous Biometadata Workshops

Info

• first attempt at very short workshop version
• alternating information about concepts and hands-on exercizes

Code of Conduct

From the The Carpentries Code of conduct:

• Use welcoming and inclusive language
• Be respectful of different viewpoints and experiences
• Gracefully accept constructive criticism
• Focus on what is best for the community
• Show courtesy and respect towards other community members

Please reach out to Maja or the helpers in case of violation of the Code of Conduct!

You will learn

• how to prepare the necessary metadata according to FAIR principles to ease the upload of your data to a public repository of your choice
• how to identify and apply minimal metadata standards
• FAIR descriptions of your datasets – Ontology terms and how to use them

What are your expectations?

What is your background with data deposition and what are your expectations? Note them in the pad.

A beginner’s guide to metadata and ontologies

Modern biology = big data?

[…] the real source of innovation in current biology is the attention paid to data handling and dissemination practices […] rather than the emergence of big data and associated methods per se. Leonelli (2016, 1)

Nothing is ever new

Is the attention to data handling and standardization really new in modern biology?

Image from SketchingScience

Model organisms

• Standardization focuses research efforts (and funds)
• But does not prohibit studying different organisms
• Quick start research on other organisms reusing findings in model organisms

Key to reusability success: curated data description and published (meta)data

Success story

Earth Microbiome Project

100 individual studies decontextualised but recontextualised for new insights!

decontextualization = using ontologies to describe a specific sample in a more broad and abstract concept recontextualization = having all necessary metadata fields to figure out the samples individual context again

Source: Thompson et al. (2017)

1. Where should you publish your data?

Data repositories

• Curated and well-described that stays on your hard drive is of limited interest to the scientific community.

• National and international efforts exist to create and maintain data repositories for the life sciences.

• For nucleotide sequence data, the INSDC integrate and mirrors data repositories from three regions:

  • USA with the NCBI

  • Europe with the EMBL-EBI

  • Japan with the DDBJ

Exercise: Data repositories

Task

• Go to https://www.ebi.ac.uk/submission
• Find out which repository could suit your data using the wizard

• Report the repository in the pad

• homework: Where do your peers usually publish their data?

• alternatively follow NFDI4Microbiota’s concise cheatsheet

Image by Andrew Martin from Pixabay

2. Depositing data means adding descriptive metadata

Metadata definition

Metadata are data about the data, or a “love note to the future” (Scott 2011).

Metadata are “reliability labels” (Leonelli 2016, 28)

Types of metadata:

• Descriptive: what is the data? e.g., title, description

• Structural: how the data is organized? e.g., file, collection

• Administrative: what is the provenance? e.g., versions, license

• Quality: How good is the data? e.g., quality rank

The categories are not fixed in marble.

Exercise: metadata

Task:

• List 1-4 metadata that you have already encountered

• Write them in the pad under one of the four types

If someone already wrote the one you thought about, add “+1” at the end!

Let’s choose ENA and figure out our required metadata

Metadata fields

• A column header expects one or more cell values
• A metadata field expects one or more values

Metadata field	Type of constraint
description	free-text
geolocation	coordinates
biome	ontology term

Metadata standards

• Fields are organized into coherent metadata standard for a given type of data

  • e.g., genomes, soil samples

• They are built and maintained by a combination of stakeholders

  • e.g., users community, data repositories

Standards expectations

Metadata standards (should) indicate for each field:

• the description of the metadata field

• the level of requirements (mandatory, recommended, optional)

• the cardinality, that is the range of expected values for the metadata field

• a persistent identifier for the field

Exercise: Metadata standards

Task:

• Go to https://www.ebi.ac.uk/ena/browser/checklists

• Look-up a standard that match your model system or type of sample

• Identify the mandatory metadata fields

• Compare with your neighbor

animal associated gut or fecal content

Minimal requirements

• The set of mandatory fields is sometimes referred to as the minimal requirements.

• Filling out these requirements and all the optional metadata fields would be ideal (if only possible) but is time-consuming

• NFDI4Microbiota started to collect overlap of minimal metadata fields across different platforms, data types and biological & environmental metadata

Exercise: Requirements

Task:

• Given only the mandatory fields, do you think you could recontextualise the data properly?
• List your arguments in the pad & note down necessary metadata fields to enhance the understanding for your dataset

ENA Browser for host-associated metadata requirements (ERC00013)

Reusability requires context & metadata

Biological context

Biology is highly contextual (few rules, many exceptions), meaning context is the key point to address to make data travel across research investigations.

Why should you share this information?

Rationale

In space, no one can hear you scream.

In data repositories, no one can hear your data description. Take the time to describe your data for humans and machines to understand!

Illustration

SAMN15720514

SAMEA103750168

• Browse to SAMN15720514
• Browse to SAMEA103750168

Environmental metadata according to established metadata standards

MIxS lists three mandatory environmental metadata fields that expect ontology terms.

Metadata field	Abbreviation	Definition
broad-scale environmental context	env_broad_scale	global correlation; ecosystem
local environmental context	env_local_scale	in local vicinity; causal influences
environmental medium	env_medium	immediate surroundings of your sample during sampling

Metadata field	Abbreviation	Recommended use of subclasses from
broad-scale environmental context	env_broad_scale	biome [ENVO:00000428]
local environmental context	env_local_scale	deeper hierarchy than broad-scale (UBERON terms accepted)
environmental medium	env_medium	environmental material [ENVO:00010483]

3. Describing human and machine-readable data

Focusing on biological and environmental metadata

What can you tell us about the biological & environmental provenance of your dataset or sample?

Where are your samples from? Do you use a model organism?

How can you make this information machine-accessible?

Why bother with ontologies?

• increase findability of your dataset
• improve machine-readability of your datasets
• help others correctly categorize & re-use your datasets > recontextualization
• required by data repositories
• first step towards open linked data and knowledge graph representations

Having fun with ontologies

Let’s talk about…

“pie”

Having fun with ontologies

• Were you thinking of the baked goods, such as cake food product [FOODON:00001278]?
• unfortunately there was no well described ontology term for pie
• Or were you thinking of the pie [44315] gene that encodes the pineapple eye protein (fruit fly) [PR:Q9VKW2]?

• add picture of a blueberry pie or blueberry cake?
• blueberry pie hyperlink to picture:
• blue berry pie picture attribution: Kelly Sue DeConnick - originally posted to Flickr as Blueberry Pie with Almond Crumb Topping licensed under Creative Commons Attribution-Share Alike 2.0 Generic

fruit fly picture attribution: Drosophila melanogaster feeding on banana by Sanjay Acharya licensed under Creative Commons Attribution-Share Alike 4.0 International

Blueberry Pie with Almond Crumb Topping by Kelly Sue DeConnick from Flickr under CC BY-SA version 2.0 Generic and Drosophila melanogaster feeding on banana by Sanjay Acharya under CC BY-SA 4.0 International

Ontology definition

• List of terms, usually taken from the scientific literature

• Ontology terms:

  • have curated textual definitions and synonyms

  • are arranged in a hierarchy from general to specific

  • have defined relationships with others terms (e.g., is_a, has_condition)

  • have persistent identifiers

  • can be cross-referenced with other resources (ontology or not)

• Ontology should reflect existing knowledge

Adapted from Osumi-Sutherland et al. (2023) and Leonelli (2016, 125, 133)

Search terms in ontologies

You can find terms in ontologies using the search bar of:

• BioPortal

• Ontology Lookup Service v4

• others: The OBO Foundry, OntoBee mainly for developers

Ontology Lookup Service

OLS is the official ontology service of the EMBL-EBI

https://www.ebi.ac.uk/ols4

• EMBL: European Molecular Biology Laboratory
• EMBL-EBI: EMBL’s European Bioinformatics Institute

Demonstration: navigating OLS for term “intestines” in UBERON, ENVO

• search intestines
• restrict search to ontologies “UBERON, ENVO”
• point out that first hit /= from requested ontology
• exact match > term not showing up
• correct spelling to “Intestine”
• explain ontology vs. present in and “imported” demonstration:
• point out: term identifier & copy function
• definition & synonyms
• term relations / class relations
• persistent identifier

Ontology definition as demonstrated with the term “intestine”

Results, we found the term

intestine [UBERON:0000160]

in both ontologies UBERON and ENVO (as imported term from UBERON)

The correct writing convention for ontology terms and their term identifiers is: > term [ontology-acronym:sequence-number], e.g. intestine [UBERON:0000160]

Exercise: Ontology browser

Task: using Ontology Lookup Service v4

• Look-up the following keywords in that order via the search bar: pond, ear and leaf

• Select a term for each using these ontologies:

  • Uber-anatomy ontology (UBERON)

  • Plant Ontology (PO)

  • Environmental Ontology (ENVO)

• Report the term and the term identifier in the pad

• pond [ENVO:00000033] (the correct term is not far down the list)

• ear [UBERON:0001690] (many “early” that can be filtered with the exact match option)

• leaf [PO:0025034] (tricky because it is an imported term)

La Grenouillère by Auguste Renoir (1869) Wikimedia

Exercise: Find your ontology terms & connect the dots

Can we start the next lesson on selecting ontology terms for YOUR dataset descriptions?

TASK: Pick any 3 terms from your dataset description.

• Note them in the pad.
• Browse the OLS and find 1-2 suitable ontology terms for each.
• Add the ontology term identifiers to the pad.
• You have 5 minutes.
• Share your experiences.
• Who used an ontology besides UBERON, ENVO and PO? And why?

Reminder: Selective ontology terms for env & biological metadata

MIxS lists three mandatory environmental metadata fields that expect ontology terms.

Metadata field	Abbreviation	Recommended use of subclasses from
broad-scale environmental context	env_broad_scale	biome [ENVO:00000428]
local environmental context	env_local_scale	deeper hierarchy than broad-scale (UBERON terms accepted)
environmental medium	env_medium	environmental material [ENVO:00010483]

Exercise: Env* metadata

Task alone or by pairs:

• Browse ENVO or UBERON (see previous table)

• List ontology terms fitting your data

• Fill out the following template on the pad:

  • env_broad_scale

  • env_local_scale

  • env_medium

• trouble finding an appropriate term downstream of the recommended class? See instructions of using other ontologies with the MIxS standard

Get more help on ENVO’s wiki page about microbial samples!

Consolidating your metadata with DataHarmonizer

• Data and metadata collection can be crucial (e.g., COVID-19)

• DataHarmonizer: a super spreadsheet to help (Gill et al. 2023)

Main features

• Load metadata standards

• Fill the template

• Validate your metadata against the template

• Easy navigation based on:
  • Sections of metadata fields
  • Metadata requirements
• Display expected values

NMDC submission portal

• National (USA) Microbiome Data Collaborative

Leverage DataHarmonizer to lower barriers to collect, study and biosample data

• Not going to use it for data submission but data description!

• Receive guidance on how to meet standards

Submission portal Demo

An otter dataset

• Sample: feces collected in the wild
• Model system: Eurasian river otter (Lutra lutra)

See the “full” methods section in the pad!

Image by Bernard-Boehne from Wikimedia Commons

Exercise/demonstration

Task:

• Go to https://data-sandbox.microbiomedata.org/submission/
• Log with your ORCID and go again to the link above
• Create a new submission using the otter dataset
• Read the methods section in the pad
• Download the XLSX template to help you validate your submission

Download the XLSX solution

your turn!

In a nutshell

Forget-me-not

• Metadata are important for re-usability of your data.

• Ontologies help scientists and machines to use common terms to help generalize your data.

Take-home message

as much as possible, as little as necessary

Working FAIRly takes time and effort¹

How much metadata is necessary to understand your research and to enable (inter-) disciplinary research?

1. 0-to-20 is already good and better than 0!

FAIR data takes time

FAIR data is a process

“Even if you don’t know how to go all the way to zero-to-60 open science, zero-to-20 is also really good”

Ellen Bledsoe in Perkel (2023)

Image by Aleš Háva from Pixabay

References

Gill, Ivan S., Emma J. Griffiths, Damion Dooley, Rhiannon Cameron, Sarah Savić Kallesøe, Nithu Sara John, Anoosha Sehar, et al. 2023. “The DataHarmonizer: A Tool for Faster Data Harmonization, Validation, Aggregation and Analysis of Pathogen Genomics Contextual Information.” Microbial Genomics 9 (1). https://doi.org/10.1099/mgen.0.000908.

Leonelli, Sabina. 2016. Data-Centric Biology: A Philosophical Study. Chicago ; London: The University of Chicago Press.

Osumi-Sutherland, David, Nicole Vasilevsky, Alex Diehl, Nico Matentzoglu, Matt Brush, Matt Yoder, Carlo Toriniai, et al. 2023. “Introduction to Ontologies.” https://oboacademy.github.io/obook/explanation/intro-to-ontologies/#key-features-of-well-structured-ontologies.

Perkel, Jeffrey M. 2023. “How to Make Your Scientific Data Accessible, Discoverable and Useful.” Nature 618 (7967): 1098–99. https://doi.org/10.1038/d41586-023-01929-7.

Scott, Jason. 2011. “The Metadata Mania.” ASCII by Jason Scott. http://ascii.textfiles.com/archives/3181.

Thompson, Luke R., Jon G. Sanders, Daniel McDonald, Amnon Amir, Joshua Ladau, Kenneth J. Locey, Robert J. Prill, et al. 2017. “A Communal Catalogue Reveals Earth’s Multiscale Microbial Diversity.” Nature 551 (7681): 457–63. https://doi.org/10.1038/nature24621.