First steps into reproducible bioinformatics

Seminar at MBIM, Nov 30, 2023

Stephan König

stephan.koenig@ubc.ca

Skylight, University of British Columbia

Mini lesson:
Reproducible reporting of computational analyses

Learning outcomes

• Describe why combining natural language and code enables reproducible analysis and research.

• Discover how symbols and text convey formatting and the structure of a document.

• Explored ways of describing microbial communities using various diversity metrics.

Microbiomes at different lake depths

Sequence barcode marker of each sample

Group similar sequences
defining
amplicon sequence variants (ASVs)
ASV ≅ species

ASV counts table

• Extract DNA from the lake samples and a sequence barcode marker.
• Group similar sequences into amplicon sequence variants, i.e., ASVs.
• ASV approximates a species.
• Record number of sequences in an ASV count table.
• Simplest way to describe is species richness.

Species richness at 10 m depth

Table 1: ASV counts

name	depth_10_m	depth_20_m
ASV_1	28	14
ASV_2	5	12
ASV_3	0	3
ASV_4	14	41
ASV_5	18	28

asv_counts |> 
  select(depth_10_m) |>
  filter(depth_10_m >= 1) |> 
  nrow()

[1] 4

• Draw a cross-section of the lake and indicate samples
• Counts for each ASV by row.
• Counts for each sample by column.
• Let’s calculate the species richness at 10 m.
• We select the corresponding column.
• An ASV is present only if the count is at least 1.
• Count of the number of rows left.

How to report the results?

• At some point you will have to report your results, for example for an assignment.
• What happens if your analysis changes? How do you keep track that the most up-to-date number is in your report?

Your Turn: Think-pair-share

How would ensure the correct reporting of your analysis?

• Think for one minute about possible solutions by yourself.
• Then discuss with your neighbour your solution.

02:00

Do you want your grades to be reproducible?

• One solution is to mix the analysis with the reporting, i.e., mix code with natural language.
• Code if often written in plain text.
• One approach is to write everything in plain text. But how can we format the text?

Your Turn: Worksheet

• Review the report of a fellow student.
• Discover how symbols and text structure the document.

Plain text

analysis.qmd

---
title: "Species richness analysis"
format: typst
---

# Identifying species in microbial communities

To describe the species found in a microbial community, we isolate the DNA of a sample and then amplify and sequence a suitable barcoding marker, typically the 16S rRNA gene. We then group all similar sequences into so-called **amplicon sequence variants (ASV)**. For our purposes here, we assume that *each ASV corresponds to a bacterial species*. By counting the number of each ASV in each sample, we get an ASV counts table (@tbl-asv-counts).

```{r}
#| label: tbl-asv-counts
#| echo: false
#| tbl-cap: "ASV counts of samples collected at different depths of a lake."

suppressMessages(library(dplyr))

asv_counts <- read.csv("asv-counts.csv")
asv_counts
```

PDF

• For our worksheet, we will act like we review the analysis of a friend.
• Your worksheet has three pages.
• The second one has a plain text file printed on a grey background called analysis.qmd.
• The third contains the PDF output of that file.
• The first one contains the exercises. We will work on the first 4.
• Please highlight any difference in text or symbols between the two files.
• Demonstrate highlighting.

Your Turn: Worksheet

• Take about 2 minute to highlight differences between plain text and the PDF output (exercise 1).
• Compare your results with your neighbour and work with them on exercise 2–4.

05:00

Wrap up: *Before proceeding, point out how students could understand some of the structure and formatting by contrasting input and output, even though it might have been challenging. You are constantly practicing a skill you will need for coding all the time: reading and understanding someone’s code.

Markdown

• Simple expressive formatting with symbols and text.
• Used in many apps and services, for example .

Markdown syntax
# Heading 1
## Heading 2
**strong**
*emphasized*

How Quarto works

• Offers many different output formats, not only PDF.
• One option is to create HTML including websites.
• In fact, this entire website was made in Quarto!

index.qmd

---
title: "MICB 2XX: Data science teaching demo"
subtitle: "Dr. Stephan Koenig"
execute:
  freeze: false
---

```{r}
#| label: setup
#| echo: false

library(conflicted)
library(dplyr)
library(fontawesome)
library(fs)

Demo MICB: 2XX website as an example of a Quarto project.

Next lab: Create your own website!

• Build a professional ePortfolio online.
• Reflect on and record your growth across the MICB program.
• Demonstrate your coding and collaboration skills.
• Showcase your work as a developing microbiologist and immunologist to future employers.

Learn relevant technologies for web development and reproducible workflows using industry-standard tools.

Universal Design for Learning:
Embracing open source

UDL strives to provide multiple means for three principles of effective instruction.

Representation

• Open source tools support alt text for images.
• Easy to provide headings and descriptive hyperlinks.
• Support for different file formats and integrating other media.

Engagement

• Follow the mastery learning model, i.e., frequent practice, assessment and feedback.
• Use online assessment platform (e.g., PrairieLearn) to support mastery learning at scale.
  • Academic integrity by randomization.
  • Use questions in practices and exams.

Engagement

• Use authentic examples and assessments (for example, ePortfolio).
• Use relevant, widespread technologies and industry standards, such as HTML, CSS, Git and GitHub, in a framework that can be adapted to any bioinformatics pipeline.
• Model inclusive practices in the way course is run and materials are provided.

Action and expression

• Alternative ways of participating, e.g., discussion section on course website.
• Increased expression and communication (open source platforms offer a lot of flexibility).
• Accommodations supported by online assessment platforms.

• Affordability is an accessibility issue.
• Uses open source material makes material more accessible.
• Cloud computing for accessibility.
• Support for desktop and mobile.
• Collect feedback on the relevant pages.
• UBC OER Fund

We are not alone: UDL community of practice

Universal design for learning fellows program—2024 cohort.

There is always more to learn from our colleagues.

Teaching philosphy: Lifelong learner

• Constructivist, thus learner centered.
• Near-limitless ability to learn.
• Raise students awareness.
• Perfectionist.
• Learning as an iterative process resembles coding to me.
• Assess students’ prior knowledge.
• Mastery learning.
• Normalizing struggle. Live coding.
• Monitor classroom climate.

Acknowlegements

• Tracy Kion and the search committee
• Hannah Wong
• Karen Smith