TUTORIAL: Comparing genomes using dotplots

Nov. 22, 2023

References:

Last - Genome Scale Sequence Comparisons

Rationale: Genomes contain enormous amounts of evolutionary information, which tell us not only about evolutionary histories of species, but also about genome organization and structure. For this reason, we need robust graphical tools that create views of the data bring out the important features of genomes. In this tutorial, we will compare a yeast genomes in the genus Saccharomyces, using the baker's yeast, S. cereviseae as the reference genome.

Goal: To discover major genome rearrangements in a pairwise comparison of two complete genomes.

This tutorial continues from the previous tutorial Finding and retrieving complete eukaryotic genomes

1. Comparison of S. cereviseae and S. arboricola genomes

The Last package contains a number of tools for comparative genomics. To create a dotplot comparing the two species, we first need to do a sequence alignment. In Last, this is done in two setps: first create a database from the reference sequence for use in later steps, and secondly, to align S. arboricola with S. cereveseae, using that database.

When working with large files, it's best to get an idea of how long various steps will take. The Unix time command can preceed any other command, and will print out a report of the time used in executing the command. To make the database, type

time lastdb -cR01 Scer GCA_000146045.2_R64_genomic.fnareal    0m5.283suser    0m4.772ssys    0m0.052s

The database will be written to a number of files with the base name Scer:

-rw-rw-r--. 1 psgendb psgendb 5842128 Nov 13 13:44 Scer.bck-rw-rw-r--. 1 psgendb psgendb      176 Nov 13 13:44 Scer.des-rw-rw-r--. 1 psgendb psgendb      509 Nov 13 13:44 Scer.prj-rw-rw-r--. 1 psgendb psgendb      136 Nov 13 13:44 Scer.sds-rw-rw-r--. 1 psgendb psgendb      136 Nov 13 13:44 Scer.ssp-rw-rw-r--. 1 psgendb psgendb 34842136 Nov 13 13:44 Scer.suf-rw-rw-r--. 1 psgendb psgendb 12071343 Nov 13 13:44 Scer.tis

The output of the time command tells us the time elapsed during the execution of the program (real), the CPU time used by the program (user) and the time required for system overhead (sys).

To create the alignment, type

time lastal Scer GCA_000292725.1_SacArb1.0_genomic.fna > ScerSarb.mafreal    0m30.982s user    0m29.582s sys    0m0.066sThis run takes about 31 sec. on our system (500 Gb RAM, 40 CPUs).

Finally, to see the results in a dotplot,

timelast-dotplot ScerSarb.maf ScerSarb.png

last-dotplot: reading alignments...last-dotplot: donelast-dotplot: choosing bp per pixel...last-dotplot: bp per pixel = 12951last-dotplot: processing alignments...last-dotplot: donereal    0m3.779suser    0m2.967ssys    0m0.084s

You can double-click on ScerSarb.png to view the output.

The X-axis sequence at top is the S. cer. genome, and the Y-axis at the left is S. arb. There are several points to note:

Each chromosome in each genome appears in the order in which they are found in the input file eg. for S. cer, they are in the roman numeral order I - XVIII plus mitochondrial at the end. Not all sequence names were able to fit in the output. However, you will see thin grey lines in the matrix plot that delineate the extents of each chromosome. For publication purposes, one could edit the genome files to say something like ScerI, ScerII etc, rather than the Accession numbers. With the default output, unfortunately, if you want to know which chromosome you are looking at, you need to count from left to right or top to bottom.
Diagonals representing matches between two chromosomes on the forward strand go top left to bottom right, and are colored red. Diagonals representing matches between two chromosomes on the reverse strands go bottom left to top right, and are shown in blue.

The plot tells us several things of importance:

The two genomes are substantially colinear, as indicated by the fact that the main diagonal spans all chromosomes.
A reciprocal translocation between Chromosomes IV and XIII is indicated by offset diagonals and gaps in the main diagonal.

It is important to keep in mind that the scale of the plots is at an extremely low level of resolution in order to view 12 Mb in a single image. What seem like minor diagonals in the plot actually span thousands of base pairs.