1 of 98

Investigating genome function in non-model mammals

Elinor K. Karlsson, Ph.D.

Associate Professor, University of Massachusetts Medical School

Director of Vertebrate Genomics, Broad Institute of MIT and Harvard

2 of 98

3 of 98

4 of 98

5 of 98

6 of 98

7 of 98

Rachel Johnston

8 of 98

Broad Institute Genomics Platform

9 of 98

Capacity today at the

Broad Institute Genomics Platform:

about one human genome every 3 minutes

10 of 98

Mersha, Tesfaye B., et al. "Resolving clinical phenotypes into endotypes in allergy: molecular and omics approaches." Clinical Reviews in Allergy & Immunology 60.2 (2021): 200-219

All of these (except proteomics & metabolomics?) are assayed with DNA sequencing technology

More sequencing capacity = we sequence more things

Also:

single cell sequencing

spatial transcriptomics

3D genome structure

11 of 98

CCGGGACGCGGCGGCCGCAGGGGCAGCGGCGACGGCAGCACCGGCGGCAGCACCGGCCGGAGCACCAGCGAGAGCAGCAGCAGCCGCGGCAGCGGCGTCCCGAGTGCCCGCGGCGCGCGGCGCAGCGATGCGGTCCCCACGGACGCGCGACCGGCCCGGGCGCCCCCTGAGCCTCCTGCTCGCCCTGCTCTGCGCCCTGCGAGCCAAGGTAGGAGCCTGCCGGGCCTCCCTCCCGGCCCGCCCTCTCCTTTCCCTCGCAGCTCCCTGGACGCCTGCGGCTGAGCGCCTCGAGCGGGCGCGGGAGCCCCCGGGCGCCCCGCTCCCCGGCTGGGTCCCCCGCGCCCGGGGGGCTGCACCTGGGCCGGGAAGTCCCGGACCTCCTACGTGTCCCCTCCACCCTCCGGGCGGGGCCGGGGGCTGTGTTCTCTCCGCTCCTCGCCCCCGGGAGCACTCGTCGGATTTCTCCGGCACTCGTGCATTTTGTGCTCGGGAATCACGTTGAGTCCTTGCACCCAGTTTTGCAAAATCCTTTTCACTCGGCGCCGGGGGCCCGTCGGGGCGGGCGGGGAGGGAGTCGCCGCTTCCACCCTCGGAGCAAACCCCTCTCCCCCGCGCTGACCTCCCTCCCCCCTCGCCGGCAGGTGTGCGGCGCCTCGGGCCAGTTCGAGCTGGAGATCCTGTCCATGCAGAACGTGAACGGGGAGCTGCAGAACGGGCACTGCTGCGGCGGCGTCCGGAGCCCGGGGGACCGCAAGTGCTCTCACGACGAGTGTGACACGTACTTCAAAGTGTGCCTCAAGGAGTACCAGTTCCGCGTCACGGCCGGGGGGCCCTGTAGCTTCGGGTCGGGCTCCACGCCAGTCCTCGGGGGCAACACCTTCAATCTCAAGGCCGGCCGTGGCAGCGAACGCAACCGCATCGTGCTGCCTTTCAGTTTCGCCTGGCCGGTGAGTGCGCCACGCGGGGAGGGCGGCCCGCCGGGGTCCCAGCCCGCGCCGGCGGAGCCCCGCGGCCTCGCCAGAGGGACGGCGGGTCTGGGCTGGAGCCTGCGCGCCGGCTGGCAAAGCCTTGCCGGGTGCGGCGTGAGGCGGCTGCGACTCCGGTTACGGTCTCCGCGGCCTCTTGCCTAGCGCGCGACAGTGGGGAGCCCGCGGGGGCTCGCGGGGGCTCGCGGGGCAAAGCTCCCAGGGAGGCGGGCTTATTAAACCTGCATCTAGAAGGCCCCAGAGTGACCCTACCCCCCCCCCCCCCCAGCTGGGCGTCTTTATGGACGATCTCTCTTTGCTTAACGAATTGAACCTGATGCGCCGTGGAAGGCGACGCGCAGTTCTGGCCTTCGAAGCCGTCCAAATGGTCACTCCCCCCTTTCTCGTGAGCTGCCGCGAGGGCGGGTGTGCCCTTCCTGGAGGGCGTGGGGGAGCCAGTTTCCCGCCGCTGCCCGGGAGACTTTGGGGCGTGCGGGGACGCGCTCCGGGCTGGACGGGAGGGTGCGGGGGCGGCCGCCGCCTGGGAGGTGGGTGGGGGTTGACTCTGGGCCGAGGCTGGAGCCGGGAGCCCGAGAGCCTCCGCCCCCTGCCGGCCCCTTCCTGCCCTCGCCCCCGCGCGGTGTAGCCCTTGTGGCCTCACTTCCAGCGGTTCCCGGGCTCTGGAACCAGCTGTGTTTGCAAACTTCCCCGGGAAGGGCGGGGGTGCGCCCTCGTCCGGCGGGCTCCGGCCCCCGCTAGCCTCCGGGGGCGTGTCAAGGCCTGGAGGGGGCGGCCCGCTGGAGGCGGCTGCGGAGAAAGTGCCAGGGCTCGGACCCCTGCCCCGGGCGCCGCTCGAGGCCCCGGGCCCGGCTGGCCCGCCCTGCAACCACCTTTCAGTTTC

?

12 of 98

3.2

billion

bases

of

DNA

13 of 98

Big question: what positions in the genome are functional / important in human health / good therapeutic targets?

14 of 98

Comparative genomics

15 of 98

16 of 98

In 100 million years, across all mammals*, all possible mutations have been tested

* placental

17 of 98

It means that any individual with that change didn’t have descendants.

If we don’t see any changes at a position in the genome, that doesn’t mean it hasn’t changed

18 of 98

branch length = average number of nucleotide substitutions per genomic position

19 of 98

	# samples	Branch length	Expected number of false positives*
29 Mammals Project	29	4.9	22,995,049
200 Mammals Project	241	16.6	191
Humans only (gnomAD)	71,702	0.17	2,604,359,690

* expected number of 100% conserved positions by random chance

20 of 98

The 200 mammals project

Resolving sequence conservation to single base pair

Kerstin Lindblad-Toh, Uppsala University & Broad Institute

And so the 200 mammals project. There are about 50 high quality placental mammalian genome assemblies out there now. We would like to sequence 150 more, yielding 200 genomes. Our power calculations indicate that this should get us to one base pair resolution for placental mammalian constraint. With this additional data, we should also be able to determine conservation information in different clades, such as primates, rodents, etc. You could imagine that for certain diseases or traits, there might be value in looking at which variants were conserved only in primates, for instance. Now annotating the human genome is our primary purpose, but in the course of this we’ll also be producing 150 genomes for 150 new species. We are collaborating with a lot of conservation scientists and ecologists and they’re all as excited as we are, since we tend to take for granted having a genome sequence – it makes a lot of what we consider to be routine work possible at all – and they were shut out from a lot of that. We’re also quadrupling the amount of data available for mammalian evolutionary biology, which is an exciting thing.

21 of 98

The 200 mammals project

placental

22 of 98

The 200 mammals project

240+

placental

23 of 98

The Zoonomia project

24 of 98

UMass Chan Medical School (Elinor Karlsson, Zhiping Weng)

Uppsala University (Kerstin Lindblad-Toh, Jennifer Meadows)

Carnegie Mellon University (Andreas Pfenning)

Earlham Institute (Wilfried Haerty)

Harvard School of Dental Medicine (Martin Nweeia)

Inst of Evolutionary Biology, Univ Pompeu Fabra (Tomas Marques-Bonet & Arcadi Navarro)

Institute for Systems Biology, Seattle (Arian Smit and Robert Hubley)

Karolinska Institite/UNC (Pat Sullivan)

Max Planck Institute (Michael Hiller)

San Diego Zoo (Oliver Ryder)

Senckenberg Research Institute, (Irnia Ruf)

ZoonomiaProject.org

Smithsonian-Mason School of Conservation (Klaus-Peter Koepfli)

Texas A & M (Bill Murphy)

Texas Tech (David Ray)

University College Dublin (Emma Teeling and Graham Hughes)

University of California San Francisco (Katie Pollard)

University of California, Davis (Harris Lewin)

University of California, Riverside (Mark Springer)

University of California, Santa Cruz (Benedict Paten, Beth Shapiro)

University of East Anglia (Federica Di Palma)

University of Nevada Las Vegas (Allyson Hindle)

University of Southern California (Steven Gazal)

University of Maine (Danielle Levesque)

Zoonomia Consortium

25 of 98

Cynthia Steiner

Aryn Wilder

Oliver Ryder

Frozen zoo: 10,000 vertebrate animals representing over 1,100 taxa, including over 200 at-risk species.

Problem: we needed to maximize species diversity

26 of 98

Cynthia Steiner

Aryn Wilder

Oliver Ryder

Frozen zoo: 10,000 vertebrate animals representing over 1,100 taxa, including over 200 at-risk species.

Problem: we needed to maximize species diversity

27 of 98

Nicole Foley

Mark Springer

Bill Murphy

Humans

Dogs

Cats

Sheep

Cattle

Horse

Goats

Mouse

Guinea pig

Ferret

28 of 98

Making genome alignment (example)

Kumar, Sudhir, and Alan Filipski. "Multiple sequence alignment: in pursuit of homologous DNA positions." Genome research 17.2 (2007): 127-135.

29 of 98

Reference-free alignment with Cactus

Benedict Paten & Joel Armstrong

30 of 98

31 of 98

… ~$100K, 2.1 million core hours, and many months later …

32 of 98

3.2

billion

bases

of

DNA

Scored conservation using PhyloP

Katie Pollard

Michael Dong

33 of 98

a

Conservation & functional scores in promoter of LDLR

phyloP

variant effect

34 of 98

Conservation & functional scores in promoter of LDLR

Zoonomia

35 of 98

more variants in humans

ClinVar

“benign”

ClinVar

“pathogenic”

deleterious

score > 20

(CADD)

More constrained across mammals

36 of 98

Cornelia Fanter

Diane Genereux

Linda Goodman

Michael Hiller

Allyson Hindle

Graham Hughes

Irene M. Kaplow

Amanda Kowalczyk

Danielle Levesque

Irina Ruf

What’s Next? Forward genomics / genotype to phenotype

Note: these are species-level or population-level phenotypes,

not individual phenotypes.

37 of 98

Forward genomics

Today

Ancient past

38 of 98

Forward genomics

Today

Ancient past

428-way human-referenced TOGA alignment of protein-coding sequences (Michael Hiller and Bogdan Kirilenko)

240-way reference-free whole genome Cactus alignment

Challenge 1: Genomics

Challenge 3: Phenotypes

Challenge 2: Methods

39 of 98

Dog genomics

40 of 98

In genomics, sample size is more important than (almost) anything else

41 of 98

42 of 98

43 of 98

44 of 98

45 of 98

We don’t know exactly which phenotype to measure to maximize power to detect genetic associations

We know more samples gives us more power

Solution: survey data

46 of 98

Useful for disease prediction

47 of 98

Useful for developing therapeutics & interventions

more detailed phenotypes required

single-cell phenotypes increasingly important

48 of 98

49 of 98

Behavior & health are a complex interaction between genetics and environment

Dogs have long been the focus of scientific intrigue. The illustration shown summarizes one of Pavlov’s classical conditioning experiments. In it, a dog conditioned to respond positively towards one image, and negatively towards another, becomes confused and distressed when shown an intermediate image. The dog demonstrates the learning process of pairing a positive stimulus or negative stimulus with neutral stimulus. But dogs are far from as simple as this behavioral paradigm might suggest.

First, domestic dogs as a model system are quite unique when compared to how we typically understand animal models. Unlike maintained laboratory animal colonies or mouse strains, and despite the genetic similarity between dogs of the same breed, dogs are still highly variable, phenotypically and genetically. From physical appearance to health and personality to fitness, the dog is a very diverse animal.

Second, dogs have a lot of the features that are ideal for a comparative animal model. They share a common environment, and even though their life experiences and perspective might differ from our own, they live richly emotional lives. They experience a lot of the same health disorders that we do, and they receive medical treatment for these disorders. Dogs and people share over 17,000 orthologous genes. Interestingly, they are an even better model than mice for the human microbiome when it comes to microbe species representation.

Third, dogs are everywhere and people love sharing information about their dogs. This is hugely advantageous for designing large, informatics-based studies. We can build open data sets more easily, less restricted by privacy concerns that face people. Research is more actionable and recruitment is more rapid. And while unfortunately our canine companions lead shorter lives than us, this is also the reason that diseases that coincide with age are easier to observe in dogs.

50 of 98

global proliferation of dogs

!

more direct human influence on selection

creation of modern breeds

1800s - present

15-20,000 years ago

domestication from ancestral wolf

Populations of dogs have been influenced by a variety of selective forces — all rooted in the human relationship — throughout their history as a species.

Domestication from ancestral wolves — wolves that no longer exist today — likely involved physiological and behavioral adaptations to anthropogenic environments. Adaptations such as improved starch metabolism and reduced fear of humans were probably beneficial for early dogs to take advantage of the resources offered by people and people’s settlements. Exactly what adaptive steps took place to produce what we might consider the first domestic dogs, we don’t know.

Today, modern wolves and modern dogs are diverged enough to be incredibly different behaviorally and physically, but genetically close enough to still be capable of hybridization.

What we do know is that following domestication, there must have been a global proliferation of dogs as a species, because, today, dogs exist nearly everywhere people do. The majority of dogs worldwide live near people but don’t live in their homes. Often, these dogs are called village or street dogs, and they are in some sense the “default” domestic dog.

What we also suspect starting happening was more direct human influence over selection, quashing troublesome behavior and promoting behaviors useful to people. For example, certain dogs that expressed attenuated hunting-like behaviors, such as stalking or heeling, might be favored, as these behaviors could be used to herd other domestic animals. Such behaviors are examples of disruptions to the typical wild type hunting sequence, a series of fixed motor patterns.

The creation of modern dog breeds, or at least the European practice of breeding dogs for show, began in the 1800s and still continues today. Breeds are isolated sub-groups of dogs bred for consistency. Many breeds may derive from lineages of dogs selected for behavior in the past, and we often think about them as representatives of different dog types.

https://www.pnas.org/doi/10.1073/pnas.2010083118

51 of 98

Health and Disease

Behavior and Disposition

Among the dog’s strengths are its immense diversity in phenotypes.

Dogs have a lot of variety in physical appearance and morphological traits, from those simply inherited like recessive brown coat to more polygenic traits controlled by multiple genes like height and pigmentation patterns.

Dogs are a model for health and disease. They can experience obesity, food allergies, and dental problems. They can develop neurological and movement disorders like degenerative myelopathy, a model of human ALS, and cancers, many of the same types that people develop.

And finally, dogs display a variety of behaviors and vary in disposition. Some dogs dig, some don’t. Some chase their tails. Some dogs are more interested in playing and others are pretty ambivalent about toys. Dogs can develop behavior problems like aggression or fear and anxiety.

52 of 98

Movies by Alice Moon-Fanelli

normal behavior done too much

distressing, time-consuming and impairing

onset in adolescence

poor response to treatment (including SSRIs)

highly heritable

Compulsive disorder in dogs

Compare 92 affected and 67 healthy dobermans

chromosome

1 gene: CDH2

Dodman et al, Molecular psychiatry 2010

53 of 98

Noh et al, Nature Communications (2017)

CDH2

unknown

54 of 98

Are dog breeds useful?

Problems:

Hard to get enough dogs in one breed

Breed comparisons are confounded by aesthetics

Regions of association are huge

Miss the (huge) power offered by admixed dogs

Advantages:

Only a sparse marker panel required

More power because of reduced genetic diversity

Breeds have characteristic phenotypes (?)

55 of 98

Open data

56 of 98

57 of 98

58 of 98

DNA extraction, sequencing, and imputation

directly to Gencove

owner activates kit and swabs dog

directly to owner

kit supplier ships to owner

owner funds

grant funds

fill out surveys on:

behavior
morphology
environment
food allergies

open canine

survey & genetic

data set

Streamlined approach to data collection

DarwinsArk.org

59 of 98

DNA extraction, sequencing, and imputation

directly to Gencove

owner activates kit and swabs dog

directly to owner

kit supplier ships to owner

owner funds

grant funds

fill out surveys on:

behavior
morphology
environment
food allergies

open canine

survey & genetic

data set

Streamlined approach to data collection

DarwinsArk.org

Brittney Logan

You can sequence your dog for only $150!!!

www.darwinsark.org

60 of 98

Ancestry-inclusive dog genomics challenges popular breed stereotypes

Morrill et al. Science. 2022.

PhD Candidate Kathleen Morrill at UMass Chan Medical

61 of 98

A community science and open data platform for studies of companion and working animal behavior and health

Darwin’s Dogs Project

Working Dog Program

Dog Cancer Project

behavior and health in companion dogs

work history and enrollment of working dogs

cancer history and enrollment for cancer projects

enrolled: 30,926 dogs

sequenced: >3,250 dogs

DarwinsArk.org

A major challenge to performing genome-wide association studies at scale is obtaining thousands of paired phenotypes and genotypes from individual dogs. That is why our lab launched Darwin’s Ark, a public science initiative, in 2015, to collect behavioral and health data on companion and working animals directly from their people. We make no restrictions on ancestry: dogs of any breed or mixed-breed ancestry are welcome. For over 30,000 dogs, we’ve collected surveys sourced from validated questionnaires on personality, impulsivity, cognitive decline, and quality of life, as well as other behaviors recommended by dog behaviorists, food allergies, and physical traits. We also gather DNA samples and have sequenced over 3,250 dogs at low-coverage with imputation of genotypes against a large panel of whole canine genomes. We recently launched two special portals to enroll working dogs and dogs with a present or past diagnosis of cancer.

62 of 98

Agonistic Threshold

Dog Sociability

Environmental Engagement

Proximity Seeking

Human Sociability

Arousal Level

Toy-directed Motor Patterns

Biddability

1

2

3

4

5

6

7

8

We identified 8 major behavioral factors

We performed exploratory factor analysis on the first 110 questions from Darwin’s Ark surveys, all addressing behavior. The first 8 factors we discovered explain 24.3% of variance in survey responses. We named these factors based on the questions captured.

Factor 1 (“Human Sociability”) captures questions about social interactions with people, especially strangers.

Factor 2 (“Arousal Level”) included questions about a dog’s reaction to excitement and impulsive behaviors.

Factor 3 (“Toy-directed Motor Patterns”) — included questions about engaging with toys and objects, which may represent underlying differences in canine motor patterns like retrieving.

Factor 4 (“Biddability”) represented the ease of training a dog and the dog’s amenability to human direction.

Factor 5 (“Agonistic Threshold”) described the conditions and contexts in which a dog may apply agonistic behavior.

Factor 6 (“Dog Sociability”) highlighted dog-directed social interactions.

Factor 7 (“Environmental Engagement”) covers questions about a dog’s responsiveness to their surroundings.

Factor 8 (“Proximity Seeking”) illustrates behaviors aimed at human contact and proximity.

63 of 98