HAL Id: hal-01938802
https://hal.archives-ouvertes.fr/hal-01938802
Submitted on 28 Nov 2018
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
The complexity ratchet: stronger than selection
Vincent Liard, Jonathan Rouzaud-Cornabas, David Parsons, Guillaume Beslon
To cite this version:
Vincent Liard, Jonathan Rouzaud-Cornabas, David Parsons, Guillaume Beslon. The complexity ratchet: stronger than selection. Beacon congress 2018, Aug 2018, east-lansing, United States. pp.1. �hal-01938802�
Abstract
Methods
Experimental design and complexity measures
Experimental design:
•
To unravel the origin of molecular complexity,
we evolved popula8ons in the simplest possible
environment: the Aevol target is a triangle.
•
We evolved 300 popula8ons of 1024 individuals
for 250,000 genera8ons under 3 muta8on rates
and monitored the evolu8on of genomic and
func8onal complexity.
The complexity ratchet: Stronger than selec;on!
Vincent Liard, Jonathan Rouzaud-Cornabas, David P. Parsons, Guillaume Beslon
INRIA-Beagle team (INSA-Lyon), Lyon, France
Using the Aevol digital gene8cs plaWorm we designed an in silico experiment to study the rela8onship between molecular complexity and phenotypic complexity: We evolved popula8ons of digital organisms in an environment designed to allow survival of the simplest possible organism: one which genome encodes a single gene. By repeatedly evolving popula8ons in this experimental framework, we observed that ≈1/3 of the lineages quickly found this simple genotype and were then stable for the rest of the experiment. At the same 8me, most lineages were not able to find this simple solu8on and showed slow gene acquisi8on along the 250,000 genera8ons of the experiment. Importantly, simple organisms ended up with a very high fitness while complex genotypes ended up with a ≈10x lower fitness. This shows that, even in a simple environment, evolu8on leads to a complexity ratchet: each gene acquisi8on creates the poten8al for the acquisi8on of further genes, ul8mately pushing evolu8on towards complex solu8ons even in a simple environment. Moreover, organisms engaged in this complexifica8on process were never able to outcompete the simple ones, showing that selec8on is not able to invert the complexity ratchet.
Aevol (www.aevol.fr) is an In Silico Experimental Evolu8on (ISEE –
aka digital gene8cs) plaWorm developed by the
Beagle team to study the evolu8on of genome structure. Aevol is based on three principles that makes it
perfectly suited to study the evolu8on of complexity:
Complexity measures:
•
Qualita8ve measure: “simple” organisms are those
encoding only proteins with the same m and w values.
•
Genomic complexity: quan8ty of informa8on encoded
on the genome (total amount of coding sequences).
•
Func8onal complexity: quan8ty of informa8on encoded
on the proteome (number of different parameters).
Functional space Activation level
m = 0.5
h = 0.5
w = 0.1
scale : 38 bp scale : 55 bpDiscussion
Results
(1) Organisms evolved complex func;onal structures in
66% of the simula;ons
Whatever the muta8on rate, ≈1/3 of the simula8ons led
to “simple” organisms with few genes and a low
func8onal complexity (A). ≈2/3 of the simula8ons led to
“complex” organisms despite the simplicity of the target
func8on (B).
(2) Complex organisms accumulate more informa;on at
the genomic and func;onal levels
Genomic complexity is strongly bounded by muta8on
rates (A) due to robustness constraints on the genome
(Knibbe et al., 2007; Fischer et al., 2014). Muta8on rates
also constrain the func8onal complexity (B) but this
effect is less stringent at the func8onal level.
(3) Simple organisms are fiKer than complex ones
Whatever the complexity measure, we observe a clear
trend for simple organisms to be fifer than complex
ones ager 250,000 genera8ons. This demonstrates that
in our simula8ons complexity is not driven by selec8on.
On the opposite, complex func8onal structures have
evolved in spite of selec8on.
(4) Despite the advantage of being simple, complex
organisms evolve greater complexity on the long term
The simple/complex iden88es are determined early on
in the simula8on and generally conserved thereager (A).
Complex organisms evolve greater complexity (B); their
fitness grows but remains far below simple organisms.
A. Genome and proteome of a simple organism B. Genome and proteome of a complex organism
A. Distribu8on of genomic complexity for complex (top) and simple (bofom) organisms. The higher the muta8on rate, the lower the genomic complexity. Genomic complexity is strongly limited by muta8onal robustness. B. Distribu8on of func8onal complexity for complex (top) and simple (bofom) organisms. The higher the muta8on rate, the lower the func8onal complexity. A. Fitness at genera8on 250,000 vs genomic complexity. The higher the genomic complexity, the lower the fitness. Simple organisms approach the op8mum fitness ( fopt = 1). Mean fitness of complex organisms: f = 0.38. A. Star8ng from simple organisms at genera8on 0, organisms’ iden8ty (simple vs complex) is determined before genera8on 10,000 and generally maintained for the rest of the simula8on. B. Long-term evolu8on of func8onal complexity in a complex organism. Func8onal complexity and fitness con8nuously grow during the 250,000 genera8ons but the fitness remains far below that of simple organisms. 0 10,000 250,000 Generation Functional Complexity 79 202 11 8 87 213 210 90 300
References
B. Fitness at genera8on 250,000 vs func8onal complexity. The higher the func8onal complexity, the lower the fitness.The emergence of complex organisms in a simple environment is a strong argument in favor of a complexity
ratchet, i.e. an irreversible mechanism that adds components to a system but that cannot get rid of exis8ng
ones, even though this could be more favorable. Indeed, in our experiments this ratchet clicks and goes on
clicking despite the selec8ve advantage of being simple. Evolu8on of fitness in complex organisms shows that
the ratchet is empowered by nega8ve epistasis. Our results show that complex biological structures can flourish
in condi8ons where complexity is not needed and that, reciprocally, the global func8on of complex structures
could very well be simple.
Experiment-specific target func8on: this triangular target func8on can be fifed by a single gene/protein. Act iva ti o n Functional space Act iva ti o n Functional space : High muta8on rate : Medium muta8on rate : Low muta8on rate : Simple organisms : Complex organisms
A. Its genotype-to-phenotype map
mimics biology with a realis8c
genomic structure and a func8onal
structure based on a graphical
formalism. Proteins are represented
by triangles which parameters are
computed from the gene sequence.
B. E v o l u 8 o n i s s i m u l a t e d b y a
genera8onal algorithm. Organisms’
fitness is based on a curve-finng task:
the protein triangles are summed to
compute the organisms’ phenotype
that is compared with a target
func8on (red curve below).
C. At each replica8on the genome may
undergo muta8ons. Aevol implements
a wide range of muta8onal operators
including switches, InDels and
chromosomal rearrangements.
Muta8ons can change complexity at
both genomic and func8onal levels.
scale : 471 bp
Genome
Proteome Phenotype
Func8onal
space Func8onal space
Ac8va8on level Ac8va8on level scale : 471 bp scale : 471 bp scale : 471 bp scale : 471 bp scale : 471 bp scale : 471 bp Chromosomal
rearrangements Switches InDels
scale : 471 bpscale : 471 bpscale : 471 bpscale : 471 bp
scale : 471 bp scale : 471 bpscale : 471 bpscale : 471 bpscale : 471 bpscale : 471 bp scale : 471 bp
Target func8on (in red)
• Liard, V., Parsons, D. P., Rouzaud-Cornabas, J. and Beslon, G. (2018) The complexity ratchet: stronger than selec8on, weaker than robustness. In: Proceedings of the 2018
Conference on Ar>ficial Life, Tokyo, July 2018, pp. 250-257.
• Knibbe, C., Coulon, A., Mazet, O., Fayard, J.-M. and Beslon, G. (2007) A long-term evolu8onary pressure on the amount of noncoding DNA. Molecular Biology and
Evolu>on, 24(10):2344-2353. • Fischer, S., Bernard, S., Beslon, G. and Knibbe, C. (2014) A model for genome size evolu8on. Bulle>n of mathema>cal biology, 76(9):2249-2291. 10 20 30 0 100000 200000 Generation Quantity of inf or mation on proteomes f = 0.031