The Cognitive Dog: Savant or Slacker

1
The Cognitive Dog Savant or Slacker

• Class 5 Observing your dog from the genome to
  the expression

2
Agenda for class...

• Questions, comments, plans for next weeks class
• Quick wrap up from last week
• The dog genome and what it tells us, or doesnt
  tell us...

3
Next weeks class...

• The origin of the dog...
• Coppinger
• Trut (optional)

4
Perception cont.
5
Smell
6
We forget that odor is a big deal

• Humans have 1000 different kinds of receptors
  (neurons) in nose for detecting odors and can
  discriminate among 10,000 different odors.
• 1 of every 100 genes in rats has something to do
  with smell...

7
A view to a nose...
Syrotuck, W. G. (1972). Scent and the Scenting
Dog. Canastota, NY, Arner Publications.
Milo D. Pearsall and H. Verbruggen (1982). Scent
Training to Track, Search and Rescue. Loveland,
CO, Alpine Publications.
8
The olfactory bulb in the dogs brain is bigger
in absolute terms than that in a human brain...
Lindsay, S. R. (2000). Applied Dog Behavior and
Training. Ames, IA, Iowa State University Press.
9
Huge difference with humans...
Humans
Dogs
Surface area of olfactory epithelium is 5 to 50x
larger in dogs, as are number of olfactory
receptors
Kaldenbach, J. (1998). K9 Scent Detection.
Calgary, Canada, Detselig Enterprises Ltd
10
Varies by breed...
Coren, S. (2004). How Dogs Think Understanding
the Canine Mind. New York, NY, Free Press.
11
What do they use their noses for?

• Prey detection
• Social functions via detection of pheromones
• Scent mark investigation (identify strange males)
• Sexual status and receptivity
• Kinship recognition
• does it bootstrap visual and auditory kin
  recognition?
• Hypothesis dogs must have layering (be able to
  keep all scents separate) and the equivalent of
  focus of attention

12
Localizing origin and direction

• .003 second difference in odor reaching one
  nostril vs. another is sufficient for determining
  general location
• Dogs seem to track by comparing difference in
  olfactory concentration between foot prints (i.e.
  discrete samples)

Higher Concentration
Lower Concentration
Bicycles are a lot harder to track because no
discrete samples
13
Taste Touch
14
Taste...

• Less sensitive than humans (especially wrt salt)
• Food preferences
• Fetal (rats seek out food eaten by mom during
  pregnancy)
• Palatable (tastes good) novel
• Prior experience (taste aversion...)

15
Touch...

• 40 of touch receptors in dogs dedicated to face
  especially around whiskers and ears
• Whiskers may help locate, identify and pick up
  small objects that might otherwise be out of
  focus.
• Distant early warning (air currents...)
• Especially sensitive between toes and pads of
  feet
• Harrington makes good point when he says that the
  role of touch is greatly under-appreciated.

16
Take home message...

• Our intuition, based on how we perceive the
  world, may not serve us well in understanding how
  your dog perceives the world.
• We think of vision as being central to our
  perception of the world
• Vision is important to dogs, but their visual
  perception may be quite different than ours, and
  olfaction plays a much bigger role, and it is
  hard for us to get our heads around olfaction
• Think movement sound

17
Take home message

• Left unsaid until now, but wicked important, and
  I do mean wicked important, is focus of
  attention perceptual bias. That is, what your
  dog attends to in any given context is likely to
  be quite different than what you attend to.
• Perceptual differences
• Emotional differences
• Motivational differences
• Learning

18
Molecular Genetics and the Dog
19
Setting the stage...
20
The archeological mystery...

• A few controversial finds dated at 12,000 -
  14,000 BP
• Is it a dog or a wolf?
• Numerous and uncontroversial finds dated at 9,000
  - 7,000 BP
• More pronounced dog-like features of skull and
  teeth

21
From Israel, 12,000 BP
Clutton-Brock, J. (1995). Origins of the dog
domestication and early history. The Domestic
Dog its evolution, behavior, and interactions
with people. J. Serpell. Cambridge, UK, Cambridge
University Press 8-20.
22
Fossil evidence

• What they are looking for...
• size of teeth
• size and proportions of skull jaw
• ...

Clutton-Brock, J. (1995). Origins of the dog
domestication and early history. The Domestic
Dog its evolution, behavior, and interactions
with people. J. Serpell. Cambridge, UK, Cambridge
University Press 8-20.
23
The proliferation of breeds is product of the
19th century

• Buffons classification of dogs 1800, today over
  150 AKC breeds

24
The most diversity of any mammalian species
Young, A. and D. Bannach (2006). Morphological
Variation in the Dog. The Dog and Its Genome. E.
A. Ostrander, U. Giger and K. Lindblad-Toh. Cold
Spring Harbor, NY, Cold Spring Harbor Press 584.

• And this doesnt even count behavioral...

25
The big questions...

• By looking at the genome of the pet dog can we
  figure out...
• The origin of the pet dog
• The evolution of breeds
• The effect of the 19th/20th century explosion in
  breeds
• The genetic basis of morphological and behavioral
  diversity

26
Quick introduction to molecular genetics...
27
What is a genome anyway?

• It describes the genetic material (DNA) within
  the nucleus of a cell.
• Sequencing the genome means creating a map that
  describes the sequence of bases on each
  chromosome.
• A gene is an identifiable part of a chromosome
  that contains the instructions to create proteins
• Most of a genome though is non-coding

Carroll, S. B. (2006). The Making of the Fittest
DNA and the Ultimate Forensic Record of
Evolutiom. New York, NY, W.W. Norton.
Proteins are a big deal because they act as
enzymes, building blocks, and help regulate
metabolism and development
28
Dogs have 39 pairs of chromosomes

• Each pair is made up of a chromosome from Mom and
  one from Dad
• For any given location on a chromosome, if it
  contains the same element as at the same location
  on its counterpart it is called homozygous, if
  different it is called heterozygous.
• Recombination and mutation create variation

Wilkie, P. J. (1999). Future Dog Breeding for
Genetic Soundness. St. Paul, MN, University of
Minnesota Agricultural Service.
29
Mitochondrial DNA

• Passed down directly from mother via egg cell.
• Only change from 1 generation to next is due to
  mutation, but rate of mutation is very low
• 500 to 1000 copies vs. 2 copies of nuclear DNA
  per cell

http//micro.magnet.fsu.edu/cells/animalcell.html
30
Microsatellites

• Repeating sequences in junk DNA at known
  locations (assumption no selective advantage)
• Mutations take the form of additions deletions
  to repeating pattern. Different lengths identify
  individuals/populations
• Change at faster rate than mtDNA

http//www.asicoaquaticmarkers.com/AnatomyofaMicro
satellite.htm
31
Practical application of microsatellite markers
Wilkie, P. J. (1999). Future Dog Breeding for
Genetic Soundness. St. Paul, MN, University of
Minnesota Agricultural Service.
32
Microsatellites
Wilkie, P. J. (1999). Future Dog Breeding for
Genetic Soundness. St. Paul, MN, University of
Minnesota Agricultural Service.

• whos my daddy?

33
Haplotypes

• Think of mtDNA (and DNA) as a long sequence of
  bases (A,T,G,C) made up of sub-sequences, known
  as Haplotypes, i.e., identifiable subsequences
• When a mutation occurs, it creates a 1 base
  change ATTA -gt ATCA. Note Prob. of base change
  at same site again is low, but needs to be
  accounted for.
• At a given region of the genome there may be one
  or more haplotypes within the population.
• e.g., 30 of the population has haplotype 1, 20
  has haplotype 2, 50 of the population has
  haplotype 3.

34
Haplotypes

• Look at...
• Number frequency of haplotypes in population
• Clustering based on closeness of different
  haplotypes
• Family tree based on model of minimum
  substitutions (one step changes)

35
Example
4 mutations away from wolf, 2 away from breed 1
ATCGAACTTTAC
breed 2
ATCGTACCTTAC
breed 1
2 mutations away from wolf
ATGGTACCTGAC
wolf
One possible interpretation breed 1 is closer to
common ancestor, and a more ancient breed...
36
Example
GAGGTATCTTAC
breed 3
ATCGAACTTTAC
breed 2
ATCGTACCTTAC
breed 1
ATGGTACCTGAC
ATGGTACCTGAC
wolf
Wolf B1 B2 B3
Wolf 0 2 4 4
B1 2 0 2 6
B2 4 2 0 8
B3 4 6 8 0
37
What does the dog genome tell us about when dogs
evolved from wolves where?
38
Big idea...

• Find parts of genetic material (DNA), at known
  locations that change randomly over time...
• Changes dont have selective advantage
• Appropriate rate of change depends on application
• Use differences between individuals to assess
• degree of relatedness
• how far back in time they diverged

39
The mitochondrial clock

• Use mtDNA as a the basis for a clock to estimate
  how long ago 2 populations diverged...
• time (amount of change)/(rate of mutation)
• e.g., 2M years (7 change)/(3.5/M year)
• Rate estimated from fossil evidence...
• 7.8 difference between coyotes and wolves
• Fossil evidence of split 1M years ago

40
Initial work in the 90s using mtDNA analysis...

• Vila et al performed a comparative analysis of
  mtDNA of dogs and wolves other canids and
  concluded...
• Dogs did seem to have evolved from wolves
• Gray wolves differ from dogs by 0.2 of mtDNA
• Wolves differ from coyotes by 4 of mtDNA
• The date of the split based on using a
  mitochondrial clock was 135,000BP
• The date was immediately called into question...

41
The mitochondrial clock...

• Problems as a clock...
• Assumes rate of mutation can be reliably
  estimated and is constant
• Across species
• Across parts of mtDNA
• Variation among lineages may not be constant
• A lineage have accumulated more or less mutations
  
• One lineage may have suffered from a collapse
  lost variation.

42
The mitochondrial clock...
bottleneck
Change that had selective advantage
lucky founder
43
The mitochondrial clock...
Multiple founding lines may produce the same
variation in less time than 1 founding line
44
Subsequent work...

• While 100,000 years is still used as the outside
  number, my interpretation is that consensus
  around 15K ypb seems to be forming in the dog
  genome community. Indeed...
• The available mtDNA data do not give resolution
  enough to precisely determine a date for the
  origin of the dog, but the archaeological record
  indicates an origin 15,000 ybp, a date which is
  not contradicted by the mtDNA record
• Savolainen, P. (2005). mtDNA Studies of the
  Origin of Dogs. The Dog and Its Genome. E.
  Ostrander, U. Giger and K. Lindblad-Toh. Cold
  Spring Harbor, NY, Cold Spring Harbor Press 584.
• Multiple domestication events, interbreeding...

45
Well, does the genome help tell us where dogs
originated?
46
More diversity, unique haplotypes in East Asia
Savolainen, P., Y.-p. Zhang, et al. (2002).
"Genetic Evidence for an East Asian Origin of
Domestic Dogs." Science 298 1610-1613.

• Is this evidence of an East Asian origin for
  dogs????

47
Ancient new world dogs...

• More closely related to old world dogs than to
  new world gray wolves.
• Seemed to be descended from 5 lineages of old
  world dogs
• Presumably introduced around 12,000- 14,000 BP
• Did not survive 2nd wave of immigration, e.g.
  todays Mexican hairless more closely related to
  old world dogs than to ancient new world dogs.

Leonard, J., R. Wayne, et al. (2002). Ancient DNA
Evidence for Old World Origin of New World Dogs.
Science. 298 1613-1616.
48
But at best this is circumstantial evidence...

• Savolainens work is suggestive but there are a
  ton of explanations for why you might see greater
  diversity in one place rather than another that
  have nothing to do with the age of the breed...
• Population size
• Bottlenecks
• Comparing apples and oranges inbred lines in
  Europe/US vs. mongrels in East Asian samples.
• Pattern of current diversity may be different
  than ancestral diversity

49
What does the dog genome tell us about breeds?
50
Parker et al used SNPs and microsatellite markers
to examine breed relationships...

• Choose to use microsatellite markers within DNA
  rather than mtDNA so as to better reflect modern
  origin of most breeds, and tried to build a
  phylogentic tree...

Parker, H., L. Kim, et al. (2004). Genetic
Structure of the Purebred Domestic Dog. Science.
304 1160-1164.
Which of these are good obedience dogs?
51
Parker et al then use statistical clustering
techniques...

• Big idea...
• Define a distance metric that allows you to say
  how close 1 pattern is to another
• Form clusters based on distance. Number of
  clusters can...
• Come from the data
• apriori, e.g., if I were to say there were 2
  clusters, find the best 2 clusters and tell me
  who would be in each

52
Pattern analysis is a very well established field

• But it does rest on apriori decisions

53
Assuming 2 clusters
Parker, H., L. Kim, et al. (2004). Genetic
Structure of the Purebred Domestic Dog. Science.
304 1160-1164.

• Note how Asian Arctic breeds, sigh hounds
  wolves are in 1 cluster

54
Assume 3 clusters
Parker, H., L. Kim, et al. (2004). Genetic
Structure of the Purebred Domestic Dog. Science.
304 1160-1164.

• 3rd cluster tend to be mastiff kinds of dogs
  broad heads lots of muscle

55
Assume 4 clusters
Parker, H., L. Kim, et al. (2004). Genetic
Structure of the Purebred Domestic Dog. Science.
304 1160-1164.

• Herding and some of the sight-hounds, but also
  shih tsu and pugs-)

56
The remaining dogs tend to be hunting dogs
terriers
Parker, H., L. Kim, et al. (2004). Genetic
Structure of the Purebred Domestic Dog. Science.
304 1160-1164.

• Herding and some of the sight-hounds, but also
  shih tsu and pugs-)

57
The technique was able to correctly assign dogs
to its correct breed in almost all cases...
Parker, H., L. Kim, et al. (2004). Genetic
Structure of the Purebred Domestic Dog. Science.
304 1160-1164.
58
Does this mean my Pharaoh hound isnt descended
from the time of Cleopatra?

• Well, its not looking good, BUT...
• The data and techniques werent sufficient to
  build a statistically strong phylogenetic tree
  for more than a few breeds. That is, it cant
  tell you who was descended from whom, and when...
• The clustering only tells how close the breeds,
  as represented by the individuals tested, are
  with respect to a certain distance metric. May
  reflect any number of things including...
• morphological similarity in part or whole
• behavioral similarity

59
And then came Tasha...

• Scientists at the Broad Institute were able to
  create a high quality map of Tashas genome
• Guide comparative analysis of human genome
• Explore genetic basis of disease susceptibility,
  morphological variation and behavioral traits

Lindblad-Toh, K., C. M. Wade, et al. (2005).
"Genome sequence, comparative analysis and
haplotype structure of the domestic dog." Nature
438(7069) 803-819.
Broad Institute
60
Some comparisons with human mouse genome

• Dog genome is 2.4GB which is smaller than human
  genome (2.9GB) or mouse genome (2.6GB)
• Dogs have 18,846 genes vs. 20426 genes in humans
  dogs may be closer to common ancestor
• High degree of synteny (similar genes line up
  together on chromosomes)
• One of the most significant findings was a
  common set of functional element corresponding
  to 5 of human genome. The stuff that makes
  organisms work?
• 1.5 protein coding genes, the remainder is
  clustered around these genes and probably include
  regulatory elements, structural elements and RNA
  genes

61
SNPs

• An SNP is a location in the genome where the
  nucleotide base (ATCG) shows some variation
  across the population.
• SNP Rate bases/SNP
• The higher the SNP, the closer the
  breeds/individuals are.

Lindblad-Toh, K., C. M. Wade, et al. (2005).
"Genome sequence, comparative analysis and
haplotype structure of the domestic dog." Nature
438(7069) 803-819.
62
Family Tree
Lindblad-Toh, K., C. M. Wade, et al. (2005).
"Genome sequence, comparative analysis and
haplotype structure of the domestic dog." Nature
438(7069) 803-819.
63
SNPs Linkage Disequilibrium

• LD is a measure that reflects the degree to which
  elements such as SNPs are correlated, i.e.,
  knowing the value of one lets you predict the
  value of another. For example,
• SNP 1 is A in 25 of the population, and SNP 2 is
  G in 10 of the population. The laws of
  probability would say that the frequency of SNP1
  A and SNP2 G in a given individual gt 2.5. If
  it is observed to be well above 2.5, then it is
  in linkage disequilibrium.
• Knowing SNP1 in this case helps predict the value
  SNP2.

64
Large homozygous blocks interspersed with small
heterozygous blocks
Lindblad-Toh, K., C. M. Wade, et al. (2005).
"Genome sequence, comparative analysis and
haplotype structure of the domestic dog." Nature
438(7069) 803-819.
65
Structure seen in Boxer consistent across breeds
Lindblad-Toh, K., C. M. Wade, et al. (2005).
"Genome sequence, comparative analysis and
haplotype structure of the domestic dog." Nature
438(7069) 803-819.
Ostrander, E. A. and R. K. Wayne (2005). "The
canine genome." Genome Research 15 1706-1716.

• But LD varies 10x across breeds reflecting origin
  pop. size.

66
The big story

• Wolf to dog created a bottleneck resulting in
  subset of wolf haplotypes
• Pre-breed dogs characterized by diverse short
  haplotype blocks
• Selective breeding created new bottleneck
• Breeds characterized by long breed specific
  haplotype blocks made up of short ancestral blocks

67
The big story continued...

• Within a typical 10KB region...
• 10 distinct haplotypes across breeds
• Within a single breed, typically see 4 haplotypes
  with the 2 most common accounting for 80 of the
  frequency
• Across breeds, haplotypes frequency vary, but a
  high degree of sharing

68
So why do you care?

• The structure of the dog genome means that it is
  dramatically easier to identify genetic basis for
  disease in dogs
• 10-15K SNPs vs. 300K SNPs in humans to provide
  coverage
• 99 chance of detecting locus given 100 affected
  and un-affected dogs in case of a single dominant
  gene
• More difficult in case of multiple interacting
  genes, but still very high chance
• Within breed and cross breed comparisons will be
  useful.

69
So why do I care?

• We arent there yet, but my guess is that in the
  next several years we will see important advances
  in our understanding of...
• The genetic basis for the incredible
  morphological diversity of dogs
• The genetic basis for behavioral perceptual
  diversity of dogs
• Remembering in both cases that what you see is a
  product of gene action X environment
• Good stuff for anyone who has dealt with canine
  lymphoma
• We probably wont see a resolution to the origins
  debate.

70
A single allele seems to be correlated with size
Sutter, N., C. Bustamante, et al. (2007). "A
Single IGF1 Allele is a Major Dterminant of Small
Size in Dogs." Science 316 112-115.