Update version of the SMBE/SESBE Lecture on ENCODE & junk DNA (Graur, December 2013)

How to Assemble a Human Genome? Mix generous amounts of Junk DNA and Indifferent DNA, add a Dollop

of Garbage DNA and a Sprinkling of Functional DNA (Lazarus DNA optional)

Dan Graur University of Houston

Dan Graur (until 5 September 2012)

Dan Graur (from 6 September 2012 to the present time)

In September 2012, 30 papers based on thousands of data sets were simultaneously published in high profile journals to describe the major findings from the ENCODE project.

The main finding of the main paper was picked up by news outlets all over the world.

And what was the main finding of the main paper?

80% of the human genome is functional

And what was the main finding of the main paper?

442 authors +

594 collaborators

On the thirtieth day of the month of September, in the Year of our Lord 2012, it was announced that “junk DNA” is “dead.”

An epic media spin

An epic media spin

An example of epic media spin An epic media spin

Compiled by T. Ryan Gregory, Genomicron

An epic media spin

An epic media spin (Una manipulación mediática épica)

Creationists had a ball

Three problems: (1) If the human genome is indeed devoid of junk DNA as implied by the ENCODE project, then a long, undirected evolutionary process, cannot explain the human genome. If, on the other hand, organisms are designed, then all DNA, or as much as possible, is expected to exhibit function. If ENCODE is right, then Evolution is wrong.

Three problems: (2) If ENCODE is right, then humans are the Goldilocks of the living world.

Organism C-value Junk Complexity Tetraodon fluvialis (pufferfish) 0.35 No Primitive Hyla nana (frog) 1.89 No Primitive Homo sapiens (human) 3.5 No Pinnacle of Creation

Extatosoma tiaratum (insect) 8.0 Yes Primitive Alium cepa (onion) 16.75 Yes Primitive Protopterus aethiopicus (lungfish) 132.83 Yes Primitive Paris japonica (canopy plant) 152.20 Yes Primitive

80% of the human genome is functional

Nature. 2011. 478:476-482

Evolutionary constraint indicates that the fraction of the human genome that is functional is ~5%.

Three problems: (3) If ENCODE-2012 is right, then ENCODE-2011 is wrong.

Solution: Kill ENCODE

We wrote a critical piece on ENCODE, and got a very negative review from Trends in Genetics. “Graur is mad, and not entirely without cause.” “It would be good for Trends in Genetics to publish a reasoned and dispassionate critical essay on this topic, preferably by someone of Graur’s stature, but not him.”

192 cm, 6’2”, 115 kg, 254 lb

angry? insane?

How did ENCODE reach the conclusion that 80% of the human genome is functional, when the evidence for selection constraint is ~5%? •  Equating hype with science. •  Wrong experimental systems. •  Inappropriate statistical analyses. •  A peculiar definition of function. •  A peculiar definition of junk. •  A lack of evolutionary perspective. •  A lack of objectivity about the study organism. •  Ignorance of everything that came before

ENCODE.

A huge chunk of ENCODE data is derived from HeLa cells and other cancer cells. Does the HeLa karyotype look human to you?

Wrong experimental systems.

Wrong experimental systems.

Landry et al. 2013

Wrong experimental systems.

•  Equating hype with science.

26

6 S E P T E M B E R 2 0 1 2

“These data enabled us to assign biochemical functions for 80% of the genome…”

The birth of 80%

27

6 S E P T E M B E R 2 0 1 2

“The vast majority (80.4%) of the human genome participates in at least one biochemical RNA- and/or chromatin-associated event in at least one cell type. Much of the genome lies close to a regulatory event: 95% of the genome lies within 8 kilobases (kb) of a DNA–protein interaction..., and 99% is within 1.7 kb of at least one of the biochemical events measured by ENCODE.”

Implication that 80% may be 99%

6 S E P T E M B E R 2 0 1 2

“junk DNA” is dead!

“junk DNA” is dead!

ENCODE researcher Ewan Birney tells Ed Yong that that the 80 percent figure will increase, possibly reaching 100 percent. “We don’t really have any large chunks of redundant DNA,” Birney says. “This metaphor of junk isn't that useful.”

99% is not enough, 100% is better

The PR machine at work: “Virtually all of the DNA passed down from generation to generation has been kept for a reason.” An intelligent God, perhaps?

I [went] back to ENCODE biologist John Stamatoyannopoulos, who was quoted in the first wave of news. He said he thought the skeptics hadn’t fully understood the papers… He did admit that the press conference mislead people by claiming that 80% of our genome was essential and useful. He puts that number at 40%. Otherwise he stands by all the ENCODE claims.

Faye Flam

99% disappears and 80% becomes 40%

40% is actually 9%

(The origin of 9%: 5% + 4% = 9%)

(Oops: 9% reverts back to 5%)

9% becomes 20%

20% kills “junk DNA”

•  20% of the genome is functional. •  Ergo, 80% must be junk. •  Yet, “junk DNA” should be “totally expunged” from the lexicon. •  In which universe does Ewan Birney’s logic work?

20% kills “junk DNA”

•  20% of the genome is functional. •  Ergo, 80% must be junk. •  Yet, “junk DNA” should be “totally expunged” from the lexicon. •  In which universe does Ewan Birney’s logic work? •  In a universe in which 20% >> 80%!

20% kills “junk DNA”

At the end of 2012, 20% becomes the favorite number in Nature

Science remains loyal to 80%.

42

Genome

Transcribed Nontranscribed

Nontranslated Translated

Information flow within the genome

43

Genome

Transcribed Nontranscribed

Nontranslated Translated

Information flow within the genome protein RNA

DNA

44 Junk Functional

Junk

Genome

Transcribed Nontranscribed

Functional

Junk Functional

Nontranslated Translated

45

Genome

Functional Junk nonfunctional

Junk has nothing to do with non-protein-coding. Junk is about function… actually lack of function.

46

Genome

Functional Junk

ad hoc

ad hoc

47

Genome

Functional Junk

= Pseudogene

48

Genome

Functional Junk

= Lazarus DNA

49

Lazarus DNA

Emmaus DNA Zombie DNA

If the acquired function (Lazarus DNA) lowers the fitness of the carriers, it is called zombie DNA.

If the acquired function (Lazarus DNA) is advantageous, it is called Emmaus DNA.

52

Transcribed Nontranscribed Transcribed Nontranscribed

Genome

Functional Junk

53

Transcribed Nontranscribed Transcribed Nontranscribed

Genome

Functional Junk

Transcriptome

Not all the transcriptome is functional.

54

Translated Untranslated

Transcribed Untranscribed Transcribed

Genome

Functional Junk

Untranscribed

Untranslated Translated

55

Translated Untranslated

Transcribed Untranscribed Transcribed

Genome

Functional Junk

Untranscribed

Untranslated Translated

Proteome Not all the proteome is functional.

THE ORIGIN OF A SPECIES (smart & elegant) Seiko Astron: Like a smartphone, the Astron is GPS-enabled, allowing it to determine accurate time from atomic clocks and automatically update to any time zone in the world. Unlike a smartphone, however, it looks nice with a suit, won’t break if you drop it and uses solar power, so it never needs to be charged. Darwin would be proud. $2,300

Hemispheres Magazine. April 2013.

This is an intelligently designed Dining Table

This is an evolutionary functional Dining Table

Evolution does not produce “smart and elegant”

Actually, Darwin would not be proud.

The story of the human genome: 1998 to the present

59

agaaacggggagaggtatcaaataaacataattgacacacccggacacgttgacttctccgttgaagttgtacgttccatgaaagttctcgacggaatagttttcatattctccgcggttgaaggtgtgcaacctcagtccgaagcaaactggagatgggcggacaggttccaagttccgaggatagccttcataaacaagatggaccgtctgggtgcggatttttacagagtgtttaaggaaatagaagaaaagctaaccataaagcccgttgccattcaaatacccctgggagcggaggaccagtttgaaggtgttatagatctaatggaaatgaaggcaataaggtggctcgaagaaaccctcggagctaaatacgaagtagtagacattcctccagaataccaggaaaaggctcaagaatggcgcgaaaagatgatagaaaccatcgtagaaaccgacgacgagttaatggaaaagtacttagaaggacaggaaatatctatagatgaactaagaaaagctttaagaaaggcaacaatagagagaaagctcgttcccgttctttgcggttctgcattcaagaacaaaggtgttcaaccccttcttgacgcagttatagattacctgccttctcctatagaccttcctcccgttaaggggacaaatcccaagaccggggaagaagaggtcagacacccctctgacgacgaacccttctgcgcttacgcctttaaggttatgtccgacccgtatgccggacaacttacctacatcagagtgttctcaggaacgctaaaagcgggttcttacgtctacaacgcaaccaaggacgaaaagcaaagggctggaagacttcttctcatgcacgcgaactccagagaggaaatacagcaggtttccgcgggtgaaatttgtgcagttgtaggactagacgccgcaacgggtgatactctctgtgatgaaaagcaccccataatccttgaaaagcttgaattccctgaccccgttatatctatggctatagagccaaagaccaagaaggaccaagaaaaactctcacaagttctcaacaagttcatgaaagaggatccaaccttcagggcaacaaccgatcccgaaactggtcagatactcatacacggaatgggtgagctccacctcgaaataatggttgacagaatgaagagggaatacggaattgaagtgaacgtcggtaaaccgcaggttgcttacaaggaaaccatcaggaaaaaggcaattggtgagggtaagttcatcaagcaaactggtggtagagggcagtacggtcacgcgataatcgaaatcgaacccctccccagaggtgcgggatttgaattcatagacgacattcacggaggagttatccccaaagaattcataccctccgttgagaagggtgtaaaggaagctatgcaaaacggaattctcgcaggataccccgttgttgacgttagagttagactctttgacggttcttaccacgaagttgactcttcggacatagcattccaggttgcgggttccttggcattcaaa 59

Bets: 281 Median: 61,302 Lowest: 25,947 Highest: 212,278 Pot: 1,200 US Dollars

The gene number game: Genesweep©

(started in Cold Spring Harbor, 1998)

Run by a Eton high-school boy called Ewan Birney. His guess was was in the very high range!

1st draft

15 February 2001

21 October 2004

From 30,000 protein-coding genes to less than 25,000.

64

agaaacggggagaggtatcaaataaacataattgacacacccggacacgttgacttctccgttgaagttgtacgttccatgaaagttctcgacggaatagttttcatattctccgcggttgaaggtgtgcaacctcagtccgaagcaaactggagatgggcggacaggttccaagttccgaggatagccttcataaacaagatggaccgtctgggtgcggatttttacagagtgtttaaggaaatagaagaaaagctaaccataaagcccgttgccattcaaatacccctgggagcggaggaccagtttgaaggtgttatagatctaatggaaatgaaggcaataaggtggctcgaagaaaccctcggagctaaatacgaagtagtagacattcctccagaataccaggaaaaggctcaagaatggcgcgaaaagatgatagaaaccatcgtagaaaccgacgacgagttaatggaaaagtacttagaaggacaggaaatatctatagatgaactaagaaaagctttaagaaaggcaacaatagagagaaagctcgttcccgttctttgcggttctgcattcaagaacaaaggtgttcaaccccttcttgacgcagttatagattacctgccttctcctatagaccttcctcccgttaaggggacaaatcccaagaccggggaagaagaggtcagacacccctctgacgacgaacccttctgcgcttacgcctttaaggttatgtccgacccgtatgccggacaacttacctacatcagagtgttctcaggaacgctaaaagcgggttcttacgtctacaacgcaaccaaggacgaaaagcaaagggctggaagacttcttctcatgcacgcgaactccagagaggaaatacagcaggtttccgcgggtgaaatttgtgcagttgtaggactagacgccgcaacgggtgatactctctgtgatgaaaagcaccccataatccttgaaaagcttgaattccctgaccccgttatatctatggctatagagccaaagaccaagaaggaccaagaaaaactctcacaagttctcaacaagttcatgaaagaggatccaaccttcagggcaacaaccgatcccgaaactggtcagatactcatacacggaatgggtgagctccacctcgaaataatggttgacagaatgaagagggaatacggaattgaagtgaacgtcggtaaaccgcaggttgcttacaaggaaaccatcaggaaaaaggcaattggtgagggtaagttcatcaagcaaactggtggtagagggcagtacggtcacgcgataatcgaaatcgaacccctccccagaggtgcgggatttgaattcatagacgacattcacggaggagttatccccaaagaattcataccctccgttgagaagggtgtaaaggaagctatgcaaaacggaattctcgcaggataccccgttgttgacgttagagttagactctttgacggttcttaccacgaagttgactcttcggacatagcattccaggttgcgggttccttggcattcaaa 64

Lee Rowen (Institute for Systems Biology) won half of the pot with a guess of 25,947 genes. She was at the bottom of the pool. Olivier Jaillon (26,500) & Paul Dear (27,462) shared the rest of the 600 dollars.

The gene number game: Genesweep©

Bets: 281 Median: 61,302 Lowest: 25,947 Highest: 212,278 Pot: 1200 US Dollars

65

Genebuild last updated/patched: !May 2012!!Total length: 3,287,209,763 bp!Protein-coding genes: 21,065!Pseudogenes: ! ! ! 15,930!RNA-specifying genes: ! 12,955!

Genebuild last updated/patched: April 2013!!Total length: 3,320,602,130 bp!Protein-coding genes: 20,774!Pseudogenes: ! ! ! 14,445!RNA-specifying genes: ! ! 22,493!

The “end” of the Human Genome Project in 2004 ($3.8 Billion) was a big disappointment for scientists unversed in evolutionary biology

The human genome turned out to be: •  small in size •  sparsely populated with genes •  densely populated with dead genomic parasites •  unoriginal

<< 3.5 billion letters in a four-letter alphabet = 7 billion bits = 0.81 GB (gigabytes)

1 DVD = 8.5 GB

Information content

Small

≈ 1 DVD = 8.5 GB

3.5 billion letters in a four-letter alphabet = = 7 billion bits = 0.81 GB (gigabytes)

Information content

Small

69

Sparsely populated with genes. Organism Gene Density

(# genes per 1 Mb)

Escherichia coli (bacterium)

911

Saccharomyces cerevisae (yeast)

483

Arabidopsis thaliana (mustard weed)

221

Drosophila melanogaster (fly)

197

Homo sapiens

12

+plus at most 0.1% for RNA-specifying genes (non-coding RNA) +plus at most 0.1% for DNA switches.

Densely populated with dead transposable elements

45-67%

Densely populated with dead transposable elements

0.15% nonsynonymous differences 1.22% synonymous differences

Unoriginal

Cost of sequencing your human genome = ~$25,000. Percent genome recovery = 90%. Error rate = 1-3%. I will provide you with your genome sequence with less error for half the price (and you can haggle).

Data: http://www.plosone.org/article/info:doi/10.1371/journal.pone.0030087

Comparing the human genome to other genomes has given rise to three complexity paradoxes.* Genomic paradox = A lack of correspondence between a measure of genome size and the presumed amount of genetic information “needed” by the organism (its complexity). *The paradoxes only exist under the assumption that humans are the most complex organisms and the pinnacle of creation.

74

Defining complexity is difficult

The complexity of a system may be defined by the minimum number of independent characters required to describe it, where independence is defined as the ability of the character to assume any possible character state independently of any other character in the system.

75

Defining complexity is difficult

Thus the wall on the right is more complex—it has a crack, than the wall on left.

Without doubt, is more complex than I

However, even if we cannot quantify organismal complexity very well, in many cases, it is possible to state unequivocally that A is more complex than B.

Without doubt, is more complex than

However, even if we cannot quantify organismal complexity very well, in many cases, it is possible to state unequivocally that A is more complex than B.

Without doubt, is more complex than I

78

K-value paradox: Complexity does not correlate with chromosome number.

46 250

Ophioglossum reticulatum Homo sapiens Lysandra atlantica

~1260

79

C-value paradox: Complexity does not correlate with genome size.

3.5 × 109 bp

Homo sapiens

6.7 × 1011 bp Amoeba dubia

1.5 × 1010 bp Allium cepa

80

G-value paradox: Complexity does not correlate with protein-coding gene number.

~21,000 ~21,000 ~57,000 >94,000

81

Total Number of Protein-Coding Genes

Drosophila melanogaster (fruitfly) 13,917 Pan troglodytes (chimpanzee) 18,746 Canis familiaris (dog) 19,856 Bos taurus (cow) 19,994 Caenorhabditis elegans (nematode) 20.517 Homo sapiens (human) 20,774 Arabidopsis thaliana (mustard weed) 27,416 Physcomitrella patens (moss) 35,938 Oryza sativa (rice) 40,577 Populus trichocarpa (poplar) 41,377 Manihot esculenta (cassava) 47,164 Malus domestica (apple) 57,386 Triticum aestivum (bread wheat) >94,000

82

Mommy, mommy, a fern has 27 times as many chromosomes

as I do; an amoeba has 200 times more DNA than I do; and wheat has 5 times more

genes than me.

83

agaaacggggagaggtatcaaataaacataattgacacacccggacacgttgacttctccgttgaagttgtacgttccatgaaagttctcgacggaatagttttcatattctccgcggttgaaggtgtgcaacctcagtccgaagcaaactggagatgggcggacaggttccaagttccgaggatagccttcataaacaagatggaccgtctgggtgcggatttttacagagtgtttaaggaaatagaagaaaagctaaccataaagcccgttgccattcaaatacccctgggagcggaggaccagtttgaaggtgttatagatctaatggaaatgaaggcaataaggtggctcgaagaaaccctcggagctaaatacgaagtagtagacattcctccagaataccaggaaaaggctcaagaatggcgcgaaaagatgatagaaaccatcgtagaaaccgacgacgagttaatggaaaagtacttagaaggacaggaaatatctatagatgaactaagaaaagctttaagaaaggcaacaatagagagaaagctcgttcccgttctttgcggttctgcattcaagaacaaaggtgttcaaccccttcttgacgcagttatagattacctgccttctcctatagaccttcctcccgttaaggggacaaatcccaagaccggggaagaagaggtcagacacccctctgacgacgaacccttctgcgcttacgcctttaaggttatgtccgacccgtatgccggacaacttacctacatcagagtgttctcaggaacgctaaaagcgggttcttacgtctacaacgcaaccaaggacgaaaagcaaagggctggaagacttcttctcatgcacgcgaactccagagaggaaatacagcaggtttccgcgggtgaaatttgtgcagttgtaggactagacgccgcaacgggtgatactctctgtgatgaaaagcaccccataatccttgaaaagcttgaattccctgaccccgttatatctatggctatagagccaaagaccaagaaggaccaagaaaaactctcacaagttctcaacaagttcatgaaagaggatccaaccttcagggcaacaaccgatcccgaaactggtcagatactcatacacggaatgggtgagctccacctcgaaataatggttgacagaatgaagagggaatacggaattgaagtgaacgtcggtaaaccgcaggttgcttacaaggaaaccatcaggaaaaaggcaattggtgagggtaagttcatcaagcaaactggtggtagagggcagtacggtcacgcgataatcgaaatcgaacccctccccagaggtgcgggatttgaattcatagacgacattcacggaggagttatccccaaagaattcataccctccgttgagaagggtgtaaaggaagctatgcaaaacggaattctcgcaggataccccgttgttgacgttagagttagactctttgacggttcttaccacgaagttgactcttcggacatagcattccaggttgcgggttccttggcattcaaagatgcagccaaaaaggcagatcccgttcttctggaacccataatggaagttgaagtggaaactcc

Conclusion: The human genome is mostly “junk.”

Ohno S. 1972. So much ‘junk’ DNA in our genome. Brookhaven Symp. Biol. 23:366-370.

?

What is and what isn’t “junk DNA”

There are known knowns; there are things we know that we know. There are known unknowns; that is to say, there are things that we now know we don’t know. But there are also unknown unknowns—there are things we do not know we don’t know.”

Donald Rumsfeld February 12, 2002

“Junk DNA” misrepresented as a “known unknown”

What is and what isn’t “junk DNA”

Junk DNA is a known known; it is a thing that we know what it does—it takes space. Junk DNA is any piece of DNA that has no function and does not affect fitness. NOT everything that is not translated or not transcribed is Junk DNA. Junk DNA is NOT a known unknown. Dark DNA is a known unknown.

Dan Graur June 22, 2013

In organisms with LARGE effective population sizes, the strength of natural selection is relatively strong.

In organisms with SMALL effective population sizes, the strength of natural selection is relatively weak.

Junk DNA is a consequence of population genetics considerations!

In organisms with LARGE effective population sizes, the strength of natural selection is relatively strong.

In organisms with SMALL effective population sizes, the strength of natural selection is relatively weak.

The majority of new mutations are mildly deleterious. In humans and elephants, selection is not sufficiently strong to eliminate many such deleterious mutations.

Junk DNA is a consequence of population genetics considerations!

In organisms with LARGE effective population sizes, the strength of natural selection is relatively strong.

In organisms with SMALL effective population sizes, the strength of natural selection is relatively weak.

Humans and elephants are expected to accumulate numerous deleterious mutations in their genome.

Junk DNA is a consequence of population genetics considerations!

Junk DNA is a consequence of population genetics considerations!

“What would we expect for the number of functional elements (as ENCODE defines them) in genomes much larger than our own? If the number [of functional elements] were to stay more-or-less constant, it would seem sensible to consider the rest of the DNA of larger genomes to be junk. If on the other hand the number of functional elements were to rise significantly with genome size, then organisms with genomes larger than ours should be more complex phenotypically than we are.”

Human exceptionalism? Genomic anthropocentrism?

W. Ford Doolittle (2013)

•  A peculiar definition of function. •  A peculiar definition of junk. •  A lack of evolutionary perspective.

In biology, there are two main concepts of function: •  A historical concept of function, also

referred to as the “selected effect function” or “proper function.”

•  A non-historical concept of function, also referred to as the “causal function.”

What is the function of the heart?

The proper function is to pump blood.

What is the function of the heart?

The proper function is to pump blood.

The causal functions of the heart are to add 300 grams to the body weight, to produce sounds, to be encased in the the pericardium, to partially fill the mediastinum, to provide an inaccurate logo for Valentine Day cards, etc.

•  Evolutionary biologists use the proper or selected effect function.

•  ENCODE used the causal function.

“Operationally, we define a functional element as a discrete genome segment that encodes a defined product (for example, protein or non-coding RNA) or displays a reproducible biochemical signature (for example, protein binding, or a specific chromatin structure).”

•  An example of a function that fits the ENCODE definition: shoes binding chewing gum.

“Operationally, we define a functional element as an entity that displays a reproducible signature (for example, chewing gum binding.”

“By the logic employed by ENCODE, following a collision between a car and a pedestrian, a car’s bonnet would be ascribed the 'function' of projecting a pedestrian many meters and the pedestrian would have the 'function' of deforming the car’s bonnet.” Laurence Hurst 2013. BMC Biol. 11:58

ENCODE uses a know logical fallacy called affirming the consequent.

If a functional sequence is transcribed,

then, all transcribed sequences are functional.

Moreover, ENCODE uses the logical fallacy inconsistently.

The ENCODE Project: 74.7% of the genome is transcribed, 56.1% is associated with modified histones, 15.2% is found in open-chromatin areas, 8.5% binds transcription factors, 4.6% consists of methylated CpGs. The fraction of the genome that is functional (the Boolean union) is 80.4%.

Our additions to ENCODE 74.7% of the genome is transcribed, 56.1% is associated with modified histones, 15.2% is found in open-chromatin areas, 8.5% binds transcription factors, 4.6% consists of methylated CpGs. 84.8% binds histone 100% of the genome is replicated. The fraction of the genome that is functional is 100%.

Interesting Question: Why do people have problems with DNA that has no function?

“… nothing is so alien to the human mind as the idea of randomness.” John Cohen. 1960. Chance, Skill, and Luck: The Psychology of Guessing and Gambling. Baltimore, MD: Penguin Books.

Inability to deal with randomness

Apophenia /æpɵˈfiːniə/: The experience of seeing meaningful patterns or connections in random or meaningless data. A type of mild or incipient schizophrenia. In statistics, apophenia is known as Type I error (false positives). Klaus Conrad. 1958. Die beginnende Schizophrenie. Versuch einer Gestaltanalyse des Wahns [Incipient Schizophrenia: An Attempt to Analyze delusion]. Stuttgart: Georg Thieme Verlag.

People like mysteries: such as hidden messages in the Bible.

If you search long enough and hard enough for patterns in random texts, you will find patterns. Especially if you do not employ negative controls. This pattern, for instance, predicts on the vertical from the bottom up (in Hebrew) that MITROMNI(TAURA)NSIA, where NSIA is “president.” The 5 letters in between MITROMNI and NSIA are random. It also helps that Hebrew has no vowels.

The Bible Code employs no negative controls. Someone else did and they found similar “prophecies” in Moby Dock by Herman Melville.

ENCODE has no negative controls. Mike White provided them and showed in a paper published in PNAS that random DNA sequences cause reproducible regulatory effects on the reporter gene. Random genetic sequences have as much or a little a function as the human genome sequences analyzed by ENCODE.

What is “junk”?

“Some years ago I noticed that there are two kinds of rubbish in the world and that most languages have different words to distinguish them. There is the rubbish we keep, which is junk, and the rubbish we throw away, which is garbage. The excess DNA in our genomes is junk, and it is there because it is harmless, as well as being useless, and because the molecular processes generating extra DNA outpace those getting rid of it.”

Sydney Brenner. 1998. Refuge of spandrels. Current Biology 8:R669.

“Were the extra DNA to become disadvantageous, it would become subject to selection, just as junk that takes up too much space, or is beginning to smell, is instantly converted to garbage by one’s wife, that excellent Darwinian instrument.”

Sydney Brenner. 1998. Refuge of spandrels. Current Biology 8:R669.

Graur’s garage: Functional but full of junk A garage in which junk became garbage

A garage according to ENCODE

Junk can Sometimes be Repurposed

Junk DNA can Sometimes be Repurposed

Norihiro Okada & Jürgen Brosius, specialists in the repurposing of junk DNA.

Functional DNA ✔ Junk DNA ✔ Garbage DNA ✔ Lazarus DNA ✔ Indifferent DNA Dark DNA

Sequence-indifferent DNA or indifferent DNA refers to DNA sites that are functional, but show no evidence of selection against point mutations. Deletion of these sites, however, are deleterious, and are subject to purifying selection.

Examples of indifferent DNA are spacers and flanking elements whose presence is required but the sequence is not important. One such case is the third position of four-fold redundant codons, which needs to be present to avoid a downstream frameshift.

Dark DNA refers to the fraction of the genome for which no good evidence exists as to its evolutionary impact on fitness. Dark DNA is an unknown unknown. The term “dark” is borrowed from the field of astrophysics.

An astrophysicist (Dr. Or Graur) whose research deals with dark energy. Unfortunately, he has no interest in dark DNA.

Interesting Question: How can one tell if a certain genomic sequence is functional or not?

Can we make the car on the left less fit for driving? Can make the car on the right less fit for driving?

119

Mutation

Evolutionary change

Mutation Mutation Mutation

Mutation Mutation Mutation

Functional DNA (almost all mutations

are deleterious)

120

Mutation

Mutation Mutation Mutation

Mutation Mutation Mutation

Nonfunctional DNA

(all mutations are neutral)

Evolutionary change

121

Since most mutations in functional regions are deleterious and likely to impair the function, these mutations will tend to be eliminated by natural selection. Thus, functional regions of the genome should evolve more slowly, and therefore be more conserved among species, than nonfunctional regions.

How do we know if a particular genomic sequence is functional?

122

Another indicator for the existence of a genomic function is that losing it has some consequence for the organism. Evolution has tested the functionality of every region of the human genome through mutation over millions of years of evolution.

123

124

Is it even possible that ENCODE is right? No! The main reason being that in humans, there is a huge difference between population size and effective population size.

Long-term Ne = 10,000

Is it even possible that ENCODE is right? Under such conditions selection is inefficient and most genetic variation is deleterious. Genomic “perfection” is unachievable.

Long-term Ne = 10,000

128

Fact 1: It has been known for more than a century that the vast majority of non-neutral mutations are deleterious (Thomas Morgan 1903). Fact 2: Mutation rate is evolvable. These facts have led Alfred Sturtevant to raise the question “Why does the mutation rate not become reduced to zero?” (Sturtevant 1937).

129

Motoo Kimura: Mutation rate cannot reach zero, because of the COST OF FIDELITY. In other words, the mutation rate in a lineage is a compromise between the benefits of complete fidelity in the replication of the genetic material and the cost of achieving complete fidelity.

The mutation rate modulation hypothesis

How did ENCODE reach such ridiculous numbers? 1.  It used methodologies encouraging biased

errors in favor of inflating estimates of functionality.

2.  It consistently and excessively favored sensitivity over specificity.

3.  It paid attention to statistical significance, rather than magnitude of the effect.

How did ENCODE reach such ridiculous numbers? 1.  It used methodologies encouraging biased

errors in favor of inflating estimates of functionality.

2.  It consistently and excessively favored sensitivity over specificity.

3.  It paid attention to statistical significance, rather than magnitude of the effect.

Example: Transcription factors binding sites (TFBS): So far, almost all known TFBSs range in length from 6 to 14 nucleotides. The TFBS entries in ENCODE range in size from 457 to 824 nucleotides. Thus, the estimates of the fraction of the human genome devoted to transcription factor bindings are extraordinarily inflated (sometimes by about two orders of magnitude).

How did ENCODE reach such ridiculous numbers? 1.  It used methodologies encouraging biased

errors in favor of inflating estimates of functionality.

2.  It consistently and excessively favored sensitivity over specificity.

3.  It paid attention to statistical significance, rather than magnitude of the effect.

Encode prefers false positives over false negatives, thus inflating the proportion of positives.

Example: ENCODE used a probability based alignment tool, and mapped RNA transcripts onto DNA when the statistical confidence exceeded 90%. This means that 10% of the correspondences between RNA and genome are erroneous. The total number of RNA transcripts in ENCODE is approximately 109 million. The mean transcript length is 564 nucleotides. Thus, a total of 6 billion nucleotides, or two times the human genome size, are potentially misplaced (false positives).

How did ENCODE reach such ridiculous numbers? 1.  It used methodologies encouraging biased

errors in favor of inflating estimates of functionality.

2.  It consistently and excessively favored sensitivity over specificity.

3.  It paid attention to statistical significance, rather than magnitude of the effect.

“Derived allele frequency spectrum for primate-specific elements, with variations outside ENCODE elements in black and variations covered by ENCODE elements in red. The increase in low-frequency alleles compared to background is indicative of negative selection occurring in the set of variants annotated by the ENCODE data.”

“Derived allele frequency spectrum for primate-specific elements, with variations outside ENCODE elements in black and variations covered by ENCODE elements in red. The increase in low-frequency alleles compared to background is indicative of negative selection occurring in the set of variants annotated by the ENCODE data.”

p = 10−37

Magnitude of effect = 0.042%

Let’s examine the rationale and the methodology for dealing with the derived allele frequency spectrum in primate-specific elements

•  If all alleles are neutral, a certain frequency distribution is expected.

•  If some alleles are under negative selection, an excess of rare derived alleles is expected.

•  This excess is expected to be detectable for only very short periods of evolutionary time.

The Why

•  To deal with very short periods of evolutionary time, ENCODE decided to use primate specific sequences.

The Why

human human human chimpanzee gorilla macaque rat mouse

human human human chimpanzee gorilla macaque rat mouse

Primate Specific Sequences

What is missing from the derived allele frequency spectrum of primate-specific elements in ENCODE? Genes! 3,296,458 SNPs that are in annotated coding regions are not found in the ENCODE sample.

Missing populations and their effect on estimates of derived alleles and ancestral alleles. Three human populations were available at the time ENCODE was submitted; ENCODE used only one.

Asia

ns

Cau

casia

ns

OUT

Derived alleles

Ancestral alleles Yo

ruba

Prim

ate

Spec

ific

Sequ

ence

s

Derived allele frequency (%) 40 60 60

Yoru

ba

OUT

Derived allele frequency (%) 100 20 0

Yoru

ba

OUT

Derived allele Frequency (%) 100 20 0

The ENCODE data includes 2,136 alleles with frequencies of exactly 0. In a miraculous feat of science, ENCODE was able to determine the frequencies of nonexistent alleles.

Frequency of derived allele = 40%

ENCODE uses multifurcated trees

Frequency of derived allele < 40%

ENCODE uses multifurcated trees

There are no derived alleles

ENCODE uses only single species from primates.

“Derived allele frequency spectrum for primate-specific elements, with variations outside ENCODE elements in black and variations covered by ENCODE elements in red. The increase in low-frequency alleles compared to background is indicative of negative selection occurring in the set of variants annotated by the ENCODE data.”

p = 10−37

Magnitude of effect = 0.042%

Unwarranted extrapolations: Badly trained techincians tend to “kill” junk DNA whenever they find a new function in non-coding DNA.

Even supposing that all the 55,000 putative lincRNAs in this paper are functional and important, then

55,000 × 2000 bp = 110 MB (less than 4% of the human genome). Showing that 4% of the genome is functional is “cool,” but doesn’t bear on the questions of “junk DNA,” which has to do with the majority of the genome.

Conclusion: Badly trained technicians who do not understand (1) population genetics, (2) the concept of effective population size, (3) random genetic drift, and (4) the limitations of selection should be forbidden to even mention “junk DNA” let alone write papers on the subject.

157

442 researchers + 288 million dollars. What have we learned from ENCODE?

6 S E P T E M B E R 2 0 1 2

“Data is not information, information is not knowledge, knowledge is not wisdom, wisdom is not truth,”

—Robert Royar (1994) paraphrasing Frank Zappa’s (1979) anadiplosis

159

The onion test is a simple reality check for anyone who thinks they have come up with a universal function for 80% of the genome, or 100% of the genome. Whatever the proposed function, ask yourself this question: Can you explain why onions need about five times more DNA than humans?”

T. Ryan Gregory

3.5 × 109 bp

Homo sapiens

1.5 × 1010 bp Allium cepa

“All science is either physics or stamp collecting.”

Ernest Rutherford “ENCODE is stamp collecting.”

Roderic Guigó

“I can think of better uses for 288 million dollars.”

Dan Graur

Acknowledgments: The Good Guys

Coauthors: Ricardo Azevedo, Becky Zufall, Nicholas Price, and Yichen Zheng (UH), and Eran Elhaik (Johns Hopkins).

Reviewers: Giddy Landan (Heirich Heine Universität, Germany), Michael Lynch (University of Indiana, USA), Naruya Saitou (National Institute of Genetics, Japan), David Penny (Massey University, New Zealand), W. Ford Doolittle (Dalhousie University, Canada + 2 reviewers who think I don’t know who they are.

Editor: Bill Martin (Genome Biology and Evolution)