About the blog author
Dr Luca Livraghi is an evolutionary-developmental biologist and Post-Doc at George Washington University in Washington DC. His work focuses on investigating genotype-phenotype relationships through butterfly wing patterns. He is a terrible cook. Follow Luca on twitter @LivraghiLuca
You are hiking through your favorite alpine trail, enjoying the breathtaking views. As you observe butterflies gracefully fluttering across the meadow, pausing to sip nectar from the dew-kissed flowers, your curiosity piques. Your attention is drawn to two remarkably similar butterflies, seemingly identical at first glance. However, a glimmering silver reflection catches you off guard. One of these winged creatures boasts silver scales adorning its wings, while the other lacks this striking feature. Intrigued, you find yourself pondering the cause of this disparity. Why do some butterflies develop silver scales while others do not? These ruminations stir a hunger within you, but it is an intellectual hunger, one that only an answer to this enigma can satiate. Fortunately, being an evolutionary-developmental biologist, you possess the perfect recipe to unravel this mystery!
Prep time: ≈20 million years | Difficulty: Off the scales | Feeds: One Post-Doc | Cooking time: HPC queues may vary
- Butterfly populations that vary for some easily scorable trait – about 30 individuals
- Butterfly nets – two
- 2mL screw caps with buffer – 60
- DNA – a few billion base pairs
- Sequencer – one
- High-performance computing cluster – one
- (De)bugs – A frustrating amount
Step 1: Finding the genes – Let Nature do the (ad)mixing for you
Thoroughly mixing ingredients is key to any good recipe. This is no different when sifting for silver genes. Quality products come from admixed populations, where similar frequencies of both silvered and unsilvered butterflies abound. Nature does the mixing for you; over many generations, mating between ancestral populations causes most of the genome to homogenize through recombination. Since we know what our trait of interest looks like, and it is an easy “yes, no” phenotype, this makes finding the underlying causal variants a lot simpler!
So, armed with this information, first find the ideal population. Then, allow the morning sun to gently simmer the butterflies awake, and collect about a dozen each. For a trait known to be under Mendelian segregation, this should be plenty.
Pro tip: Remember to take note of each of the butterflies’ phenotypes!
Step 2: Mash, extract, and sequence!
Once you have gathered your ingredients, it is time to mash! To arrive at the juicy candidate gene, use a piece of thorax from each collected butterfly, and extract about 3µg of DNA per individual. Make sure to clean up the DNA (you don’t want any clumpy bits in your sequencer) and attach your favourite barcodes (I always like the Illumina flavors for their crispy demultiplexing*). Then incubate in your favourite sequencer for at least 8 hours. Wait for the DING (or whatever noise sequencing machines make these days) and voila, your chopped-up pieces of DNA are ready for analysis!
*The author of this recipe is not sponsored by Illumina but is currently taking offers.
Step 3: Bake in a high-performance computing cluster
For this step, you will need a pre-assembled genome for your species of interest. Luckily, long-read sequencing technology has made this step quicker and cheaper than ever (PacBio, I also love you!). You now want to map your bits of DNA for each individual collected, back to your pre-assembled genome. Once this step is completed, you are ready to call your genotypes! You can now write your favorite gatk bash script and are ready to get frustrated at HPC queue times.
Pro tip: You know that debug partition on the cluster that is meant to be reserved just for, ehm, debugging? This author sees that more as a guideline than a strict rule.
Step 4: The Manhattan Plot: an effervescent delight
If you haven’t Wall-E’d your laptop in frustration by this point, you are now ready to plot! Each of your butterflies will have been assigned a genotypic flavor for each position along the genome. For the second round of baking, your bash script will now logistically regress each of these flavors against your recorded phenotype (you didn’t forget to score these did you!?). In non-chef parlance, you are asking if your genotypic state is correlated with the phenotype, for each point in your genome. Cook until plink, and wait for the p-values to evaporate out. If you’ve done this step correctly, you should now see a single stream of effervescent bubbles start emerging from one chromosome*. You now have your candidate locus!
*Only applies to the Mendelian trait recipe.
Step 5: Smash it with a hammer
We will now employ the sophisticated culinary technique of smashing things up and seeing what happens. Your beautiful gassy peak should have hopefully pointed you to an obvious candidate gene (it’s a Wnt ligand isn’t it)? Use your favorite genetic manipulation tool (CRISPR is recommended), and go ahead and break that gene. Incubate for about a month (butterfly development times may vary). Your butterfly should have now emerged from its chrysalis, finally the product is ready. Serve with a side of “how this fits in a wider GRN” hypothesis, and display your creation to your eagerly awaiting journals.
Pro tip: Critic Number 2 will no doubt ask to confirm with qPCR. Feel free to ignore.