molecular marker

From genetic to genomic markers

As we discussed in last week’s post, different parts of the DNA can be used as genetic markers for analyses relating to conservation, ecology and evolution. We looked at a few different types of markers (allozymes, microsatellites, mitochondrial DNA) and why different markers are good for different things. This week, we’ll focus on the much grander and more modern state of genomics; that is, using DNA markers that are often thousands of genes big!

Genomics vs genetics — If we pretended that the size of the text for each marker was indicative of how big the data is, this figure would probably be about a 1000x under-estimation of genomic datasets. There is not enough room on the blog page to actually capture this.

I briefly mentioned last week that the development of genomics was largely facilitated by what we call ‘next-generation sequencing’, which allows us to easily obtain billions of fragments of DNA and collate them into a useful dataset. Most genomic technologies differ based on how they fragment the DNA for sequencing and how the data is processed.

While the analytical, monetary and time cost of obtaining genomic data has decreased as sequencing technology has improved, we still need to balance these factors together when deciding which method to use. Many methods allow us to put many individual samples together in the same reaction (we tell which sequence belongs to which sequence using special ‘barcode sequences’ that code for one specific sample): in this case, we also need to consider how many samples to place together (“multiplex”).

As a broad generalisation, we can separate most genomic sequencing methods into two broad categories: whole genome or reduced-representation. As the name suggests, whole genome sequencing involves collecting the entire genome of the individuals we use, although this is generally very expensive and can only be done with a limited number of samples at a time. If we want to have a much larger dataset, often we’ll use reduced-representation methods: these involve breaking down the whole genome into much smaller fragments and as many of these as we can to get a broad overview of the genome. Reduced-representation methods are much cheaper and are appropriate for larger sample sizes than whole genome, but naturally lose large amounts of information from the genome.

Genomic sequencing pathway — The (very, very) vague outline of genomic sequencing. First we take all of the DNA of an organism, breaking it into smaller fragments in this case using a restriction enzyme (see below). We then amplify these fragments, making billions of copies of them before piecing them back together to either make the entire genome (left) of a few individuals or patches of the genome (right) for more individuals.

Restriction-site associated DNA (RADseq)

Within the Molecular Ecology Lab, we predominantly use a technology known as “double digest restriction site-associated DNA sequencing”, which is a huge mouthful so we just call it ‘ddRAD’. This sounds incredibly complicated, but (as far as sequencing methods go, anyway) is actually relatively simple. We take the genome of a sample, and then using particular enzymes (called ‘restriction enzymes’), we break the genome randomly down into small fragments (usually up to 200 bases long, after we filter it). We then attach a specific barcode for that individual, and a few more bits and pieces as part of the sequencing process, and then pool them together. This pool (a “library”) is sent off to a facility to be run through a sequencing machine and produce the data we work with. The ‘dd’ part of ‘ddRAD’ just means that a pair of restriction enzymes are used in this method, instead of just one (it’s a lot cleaner and more efficient).

ddRAD flowchart — A simplified standard ddRAD protocol. 1) We obtain the DNA-containing tissue of the organism we want to study, such as blood, skin or muscle samples. 2) We extract all of the genomic DNA from the tissue sample, making sure we have good quantity and quality (avoiding degradation if possible). 3) We break the genome down into smaller fragments using restriction enzymes, which cut at certain places (orange and green marks on the top line). We then attach special sequences to these fragments, such as the adapter (needed for the sequencer to work) and the barcode for that specific individual organism (the green bar). 4) We amplify the fragments, generating billions of copies of each of them. 5) We send these off to a sequencing facility to read the DNA sequence of these fragments (often outsourced to a private institution). 6) We get back a massive file containing all of the different sequences for all of the organisms in one file. 7) We separate out these sequences into the individual the came from by using their special barcode as an identifier (the coloured codes). 8) We then process this data to make sure it’s of the best quality possible, including removing sequences that we don’t have enough copies of or have errors. From this, we produce a final dataset, often with one continuous sequence for each individual. If this dataset doesn’t meet our standards for quality or quantity, we go back and try new filtering parameters.

Gene expression and transcriptomics

Sometimes, however, we might not even want to look at the exact DNA sequence. You might remember in an earlier blog post that I mentioned genes can be ‘switched on’ or ‘switched off’ by activator or repressor proteins. Well, because of this, we can have the exact same genes act in different ways depending on the environment. This is most observable in tissue development: although all of the cells of all of your organs have the exact same genome, the control of gene expression changes what genes are active and thus the physiology of the organ. We might also have genes which are only active in an organism under certain conditions, like heat shock proteins under hot conditions.

This can be an important part of evolution as being able to easily change genetic expression may allow an individual to adapt to new environmental pressures much more easily; we call this ‘phenotypic plasticity’. In this case, instead of sequencing the DNA, we might want to look at which genes are expressed, or how much they are expressed, in different conditions or populations: this is called ‘comparative transcriptomics’. So instead of sequencing the DNA, we sequence the RNA of an organism (the middle step of making proteins, so most RNAs are only present if the gene is being expressed).

Processing data

Despite how it must appear, most of the work with genomic datasets actually comes after you get the sequences back. Because of the nature and scale of genomic datasets, rigorous analytical pipelines are needed to manage and filter data from the billions of small sequences into full sequences of high quality. There are many different ways to do this, and usually involves playing with parameters, so I won’t delve into the details (although some of it is explained in the boxed part of the flowchart figure).

The future of genomics

No doubt as the technology improves, whole genome sequencing will become progressively more feasible for more species, opening up the doors for a new avalanche of data and possibilities. In any case, we’ve come a long way since the first whole genome (for Haemophilus influenzae) in 1995 and the construction of the whole human genome in 2003.

What is a ‘molecular marker’?

As we’ve previously discussed within The G-CAT, information from the DNA of organisms can be used in a variety of ways to study evolution and ecology, inform conservation management, and understand the diversity of life on Earth. We’ve also had a look at the general background of the DNA itself, and some of the different parts of the genome. What we haven’t discussed yet is how we use the DNA sequence in these studies; most importantly, which part of the genome to use.

The genome of most organisms is massive. The size of the genome ranges depending on the organism, with one of the smallest recorded genomes belonging to a bacteria (Carsonella ruddi), consisting of 160,000 bases. There is a bit of debate about the largest recorded genome, but one contender (the ‘canopy plant’, Paris japonica) has a genome stretching 150 billion base pairs long! The human genome sits in the middle at around 3 billion bases long. Naturally, it would be incredibly difficult to obtain the sequence of the whole genome of many organisms (particularly 20 – 30 years ago, due to technological limitations in the sequencing process) so instead we usually pick a specific region of the genome instead. The exact region (or type of region) we use is referred to as a ‘molecular marker’.

How do we choose a good marker?

The marker we pick is incredibly important: this is often based on how much variation we need to observe across groups. For example, if we want to study differences between individuals, say in a pedigree analysis, we need to pick a section of the DNA that will show differences between individuals; it will need to mutate fairly rapidly to be useful. If it mutates too slowly, all individuals will look identical genetically and we won’t have learnt anything new at all.

On the flipside, if we want to study evolution at a larger scale (say, between species, or groups of species) we would need to use a marker that evolves much slower. Using a rapidly mutating section of DNA would effectively give a tonne of ‘white noise’; it’d be impossible to pick what is the genetic difference at the species level (i.e. one species is different to another at that base) vs. at the individual level (i.e. one or many individuals within the species are different). Thus, we tend to use much slower mutating markers for deeper evolutionary history.

Evol spectrum — The spectrum of evolutionary history, with evolutionary splits between major animal groups on the left, to splits between species in the middle, to splits between individuals within a family tree on the right. The effectiveness of a marker for a particular part of the spectrum depends on its mutation rate. The original figure was taken from a landmark paper by Avise (1994), considered one of the forefathers of molecular ecology.

Think of it like comparing cats and dogs. If we wanted to compare different cats to one another (say different breeds) we could use hair length or coat colour as a useful trait. Since some breeds have different coat characteristics, and these don’t vary as much within the breed as across breeds, we can easily determine a long haired cat from a short haired cat. However, if we tried to use coat colour and length to compare cats and dogs we’d be stumped, because both species have lots of variation in these traits within their species. Some cats have coat length more similar to some dogs than to other cats for example; so they’re not a good characteristics to separate the two animal species (we might use muzzle shape, or body shape instead). If we substitute each of these traits with a particular marker, then we can see that some markers are better for some comparisons but not good for others.

Allozymes

The most traditional molecular marker are referred to as ‘allozymes’; instead of comparing actual genetic sequences (something that was not readily possible early in the field), variations in the shape (i.e. the amino acids of the protein, not the code underlying it) were compared between species. Changes in proteins occur very rarely as natural selection tends to push against randomly changing protein structure, since the shape of it is critical to its function and functionality. Because of this, allozymes were only really effective for studying very broad comparisons (mainly across species or species groups); the exact protein used depends on the study organism. Allozymes are generally considered outdated in the field nowadays.

With the development of technologies that allowed us to actual determine the DNA code of genes, molecular ecology moved into comparing actual sequences across individuals. However, early sequencing technology could generally only accurately determine small sections of DNA at a time, so particular markers capitalising on this were developed. Many of these are still used due to their cost-effectiveness and general ease of analysing.

Microsatellites

For comparing closely related individuals (within a pedigree, or a population), markers called ‘microsatellites’ are widely used. These are small sections of the genome which have repetitive DNA codes; usually, the same two or three base pairs (one ‘motif’) are repeated a number of times afterwards (the ‘repeat number’). While the motifs themselves rarely get mutations, the number of repeated motifs very rapidly mutates. This is because the protein that copies DNA is not very perfect, and often ‘slips up’, and adds or cuts off a repeat from the microsatellite sequence. Thus, differences in the repeat number of microsatellites accumulate pretty quickly, to the point where you can determine the parents of an individual with them.

Microsat_diagram — The general (and simplified) structure of a microsatellite marker.

Microsatellites are often used in comparisons across closely related individuals, such as within pedigrees or within populations. While they are relatively easy to obtain, one drawback is that you need to have some understanding of the exact microsatellite you wish to analyse before you start; you need to make a specific ‘primer’ sequence to be able to get the right marker, as some may not be informative in particular species or comparisons. Many researchers choose to use 10-20 different microsatellite markers together in these types of studies, such as in human parentage analyses.

Cats_parentage — Microsatellites are useful for parentage analysis. Our previous guest contestants are here to discuss ‘Who is the father?!’ in *Maury*-like fashion. The results are in, and using 4 microsatellites (1-4) and looking at the number of repeats in each of those, we can see the contestant 2 is undoubtedly the father! I’ll be honest, I have no idea if this is how *Maury* works, but I think it would work.

Mitochondrial DNA

For deeper comparisons, however, microsatellites mutate far too rapidly to be effective. Instead, we can choose to use the DNA of the mitochondria. You may remember the mitochondria as ‘the powerhouse of the cell’; while this is true, it also has a lot of other unique properties. The mitochondria was actually (a very, very, very long time ago) a separate bacteria-like organism which became symbiotically embedded within another cell. Because of this, and despite a couple billion years of evolution since that time, the mitochondria actually has its own genome separate to the ‘host’ (like the standard human genome). The full mitochondrial genome consists of around 37 different genes, most of which don’t code for any proteins involved directly in evolution; as such, natural selection doesn’t affect them as much as other genes. The most commonly used mitochondrial genes are the cytochrome b oxidase gene (cytb for short) or the cytochrome c oxidase 1 (CO1) gene.

The mitochondrial genome evolves relatively rapidly (but not nearly as fast as microsatellites) and is found in pretty much every plant and animal on the planet. Because of these traits, it’s often used as a way of diagnosing species through the ‘Barcode of Life’ project (using cytb and CO1). It’s very widely used within species-level studies, to the point where we can even use the relatively consistent mutation rate of the mitochondrial genome to estimate how long ago different species separated in evolution.

Cats_barcode — Not entirely how the *Barcode of Life* works, but close enough, right?

Other markers?

There are plenty of other genetic markers that are used within molecular ecology, with some focusing on only the exons or introns of genes, or other repetitive sequences. However, microsatellites and mitochondrial genes are among the most widely used in evolution and conservation studies.

While these markers have been very useful in building the foundations of molecular ecology as a scientific field, developments in sequencing technology, analytical methods and evolutionary theory have pushed our ability to use DNA to understand evolution and conservation even further. Particularly the development of sequencing machines which can process much larger amounts of genetic DNA. This has pushed genetics into the age of ‘genomics’: while this sounds like a massively technical difference, it’s really just about the difference in the size of the data we can use. Obviously, this has many other benefits for the kinds of questions we can ask about evolution, conservation and ecology.

Genomics has massively expanded in recent years, the types, quantity and quality of data are diverse. Stay tuned because next week, we’ll start to delve into the modern world of genomics!