Skip to content

Latest commit

 

History

History
378 lines (211 loc) · 7.19 KB

lecture_6.md

File metadata and controls

378 lines (211 loc) · 7.19 KB
title author institute date output
Pangenome visualisation
Alexander Leonard
ETH Zürich
Day 2
beamer_presentation
theme keep_tex
Boadilla
true

Recap

We have learned about:

  • how accurate long reads have reshaped genome assembly
  • pangenomes are a natural step to encapsulating all the new genomes
  • There is no "one true pangenome", but each has their own strengths

Objectives

By the end of the lecture, we should be able to:

  • create figures of pangenomic bubbles
  • interactively explore pangenomes with different layers of data
  • identify large-scale homology between pangenome assemblies

Overview

  • local-scale visualisations
  • global-scale visualisations
  • layering visualisations with additional data

Visualising genomic data

What types of genomic data do we normally try and visualise?

Visualising genomic data

IGV (Integrative Genomics Viewer, https://igv.org/doc/desktop/) is a useful tool for visualising different formats of genomic data:

  • read alignments
  • bed files
  • gene annotations

. . .

Seeing the data can often influence later analyses:

  • too many/few reads where we expect them
  • overlap of variants and complex annotations

Visualising genomic data

IGV example{ width=100% }

Visualising genomic data

There are many other ways to visualise genomic data, such as:

  • JBrowse
  • Ribbon
  • USCS Genome Browser

. . .

Is there a pangenomic equivalent?

Visualising pangenomic data

Sadly, not really...
Everything is more complicated in the pangenomic world.

. . .

But it depends what are we interested in:

  • viewing relationship between many assemblies?
  • viewing alignments/annotations on pangenome graphs?

Visualising pangenomic data

How do we visualise the GFA output of pangenome construction?

One of the most common tools is Bandage (https://github.com/asl/BandageNG).
It has several advantages:

  • easy to install
  • quick to load small-moderate graphs
  • lots of extra functionality

Bandage visualisations

A relatively easy example of a minigraph bubble.

minigraph example{ width=80% }

Bandage visualisations

A not-so-easy example of that region in cactus.

cactus example{ width=80% }

Bandage visualisations

A hard example of that region in pggb.

pggb example{ width=30% }

Using Bandage interactively

Beyond viewing graphs, we can also use Bandage for:

  • searching for sequence hits (blastn, minimap2, etc.)
  • annotating paths
  • loading BED files

Using Bandage interactively

Let's explore Bandage a bit.

Beyond bubbles

Bandage is a powerful tool for working on a local scale.

How can we look at pangenomes (and the relationships between assemblies) on a global scale?

Pangenomic synteny

Synteny Circos plots can be an informative way to compare assemblies.

. . .

cicros plot{ width=50% }

Pangenomic synteny

We can easily construct "multi-assembly" synteny plots.

. . .

cicros plot{ width=50% }

. . .

But are they helpful?

Pangenomic synteny

Many one-to-one alignments is not the same as many-to-many alignments.

. . .

Very easy to misinterpret or even miss key relationships.
But, this can be a helpful stepping stone to transition to pangenomic concepts.

Pangenomic synteny

Viewing too much information can be just as unhelpful as viewing too little.

pafplot{ width=50% }

Pangenomic synteny

pafplot zoom{ width=50% }

Even the variation you see is hard to relate amongst all assemblies.

A quick break

And then some odgi!

Pangenomic synteny

Another critical pangenome tool is odgi (https://github.com/pangenome/odgi).

odgi is a play on the Italian word "oggi" (/ˈɔd.dʒi/), which means "today".
As of 2019, a standard refrain in genomics is that genome graphs will be useful in x years.
But, if we make them efficient and scalable, they will be useful today.

We can use odgi viz to get something in between 1D linear (easy) and pangenome (informative) views.
This bins the pangenome and produces a linear, static visualisation of the graph.

odgi visualisations

odgi depth{ width=95% }

. . .

Nodes are "ordered" left to right, but what does that mean?

. . .

THEY ARE NOT NECESSARILY SEQUENTIAL

. . .

How do we interpret the links (graph topology)?

odgi visualisations

We can also plot a "compressed" mode, and see which regions are variable.

odgi compressed{ width=95% }

odgi visualisations

There are many additional layers of information we can use:

  • inversions
  • traversal depth
  • any annotations from BED files?

odgi inject

We can "inject" an annotation from any assembly into the pangenome.

. . .

odgi inject{ width=95% }

. . .

Let the pangenome do the hard work for dealing with pangenome coordinates!

odgi untangle

We can also "untangle" a graph bubble locally, and "linearise" it.

odgi tangle{ width=90% }

odgi untangle

Easier to see copy number variation (like VNTRs) or gene structure.

. . .

odgi untangle{ width=50% }

So linear visualisations in the end?

Pangenomes are efficient data structures.

. . .

Pangenomes can be efficient visualiation structures.

. . .

Linear sequence context will always make more sense to us.

. . .

These plots still use pangenomics before the final image.

Pangenome visualisations

These tools are also likely best used in combination, so we can understand the graph at different scales.

multiple visualisers{ width=30% }

. . .

"Publication figures" might focus on different "graphness" levels.

Other pangenome visualisers

There are many other tools:

  • Waragraph (and earlier gfaestus; depends on odgi layout)
  • sequenceTubeMap from vg
  • Panache
  • Panagram

. . .

And even more "Pan" puns at https://github.com/colindaven/awesome-pangenomes.

Summary

. . .

:::incremental

  • Visualising your data is critical, even more so for pangenomes!
  • Bandage is powerful for interactively exploring "local" regions.
    • Increasing graph complexity will be impossible to responsively display.
  • odgi is powerful for statically visualising entire graphs. :::

Practical: Hands on pangenome assembly of a chromosome.

Goals of this afternoon.

Part 1:

  • assemble an entire chromosome from long reads
  • build a chromosome pangenome from six assemblies

Part 2:

  • visually explore minigraph and pggb pangenomes
  • examine sequences and annotated features in pangenomes

Questions?

And then coffee