Affinity-based DNA Synthetic Evolution (ADSE). a: structure of the DNA oligomers participating in ADSE as individuals (DNAi) and resources (Figure 1-figure supplement 1). b: steps in the ADSE (Figure 1-figure supplement 2). The process starts with a random-sequence DNAi population. The capture by magnetic bead-conjugated resources provides the selection: bead-bound DNAi are amplified to form the new generation, a small fraction of which is sequenced by MPS. The rest of the original solution is discarded. Red arrows mark the steps of each ADSE cycle. c: possible interactions involving DNAi. The online version of this article includes the following figure supplement(s) for Figure 1: Figure 1-figure supplement 1 DNA individuals structure and its interaction with the oligos. Figure 1-figure supplement 2 Schematic representation of the Affinity-based DNA Synthetic Evolution protocol.

Evolution of the DNAi population. Time is expressed in ADSE cycles. a: fraction FD of the total population formed by different sequences obtained from experiment (red dots) and computed with the IBEE model (green line); fraction F10 of the total population formed by the 10 most abundant sequences (experimental: black dots, IBEE model: grey line). c: boxplots and scatterplots of ⟨ ΔGDR⟩ in ensembles of 1000 random sequences (left), 1000 sequences extracted from the experimental population at cycle 24 (middle) and top-10 most populous sequences at the same cycle (right). The color code is assigned to each point based on its ω value (color bar). d-f: probability distributions P (ω) for at cycles 0 (d), 12 (e) and 24 (f). The orange points and line are the distributions evaluated with the null model. g-i: evolution of the abundance (expressed as fraction of the total population F) of sequences whose ω is low (3 ≤ ω ≤ 5-panel g), medium (6 ≤ ω ≤ 10 - panel h) and high (11 ≤ ω - panel i) as obtained from the experiments (dots) and with the IBEE model (green lines). Model results are averages over 20 simulations. The online version of this article includes the following figure supplement(s) for Figure 2: Figure 2-figure supplement 1. Time evolution of the fraction of PCR by-products and of their ⟨ω⟩. Figure 2-figure supplement 2. Probability distributions P (⟨ω⟩) of PCR by-products at cycles 0 and 24. Figure 2-figure supplement 3. Time evolution of ⟨ ω ⟩for the replica Oligo2. Figure 2-figure supplement 4. Probability distributions P (⟨ω⟩) at cycles 0 and 18 for the replica Oligo2. Figure 2-figure supplement 5. Experimental vs IBEE time evolution of the population zip ratio and of the Shannon Entropy associated to the RSA distribution. Figure 2-figure supplement 6. Experimental vs IBEE ⟨ω⟩ time evolution, for two different IBEE hyperparameters choices. Figure 2-figure supplement 7. Experimental vs IBEE ⟨ω⟩ time evolution, for different IBEE starting population sizes and compositions.

Distribution and evolution of free energy quantifiers. a: probability distribution for the DNAi-resource binding free energy computed by NUPACK, p(⟨ΔGDR⟩), for the initial population (grey shading), for the final population (random choice of 1000 DNAi - cyan columns, top 10 most populous sequences - black columns). b: time evolution, expressed in cycles, for various mean free energies⟨ΔG⟩, normalized to their value computed on pools of random sequences. All ⟨ΔG⟩ are computed by NUPACK on sets of 1000 individuals: ΔGDR (blue dots); unimolecular self interaction ⟨ΔGself⟩ (green dots); bimolecular mutual DNAi interaction ⟨ΔGDD⟩ (red dots); mutual, self-subtracted interaction ⟨ΔΔG⟩ (yellow dots). c: scatter plot of ⟨ΔGself⟩ vs. ΔGDR computed for 1000 DNAi in the final population (blue squares). Red dots mark the point relative to the 10 most populous sequences, as identified by the labels. Note that the x-axis scale of panels a and c is the same, enabling identifying sequences. d: scatter plot computed on 104 DNAi pairs from the final population comparing ΔGselfj + ΔGself,l and ΔGDD,kl (green squares). Red dots mark the pair formed by the 10 most populous sequences, some of which identified by labels. With respect to the condition ΔGDD,kl = ⟨ ΔGself⟩,j + ⟨ ΔGself⟩,l (black line), data are on average displaced by ΔΔG ∼ 7.5 kcal/mol (yellow arrow). The online version of this article includes the following figure supplement(s) for Figure 3: Figure 3-figure supplement 1. Three examples of different relative positions for sequence-sequence attachment. Figure 3-figure supplement 2. Scheme of target-consumer interaction, from a combinatorial standpoint. Figure 3-figure supplement 3. Null model without threshold, analytical and simulated, with an even and uneven nucleotides distribution. Figure 3-figure supplement 4. Null model with/without threshold, simulated and analytical.

Natural history of DNAi species. a: fraction F of {DNAi} that belong to a choice of specific species as a function of the ADSE cycles, in linear (left) and logarithmic (right) scale. Arrows connect the initial condition (one individual per species) to the earliest detection via sequencing, across the 6 order of magnitudes gap (grey shading) between the population of DNAi survived at each cycle and the amount of sequenced DNAi. The same growth is assumed for species 12.1, suggesting its appearance by mutation occurred at generation 9. b-f: self interactions (b1, c1, e1), resource interactions (b2, c2, e2, d1, d2) and mutual interactions (b3, c3, e3, d1, d2, f) of selected species, sketched as per the NUPACK output. Nucleobases color coded (G -black, C - blue, A - green, T - red). Paired bases are connected. Double and single stand regions are represented as straight and curved, respectively. As in Figure 1a, terminal blocks of DNAi are marked as graphic double helices colored according to the legend of panel a, and beads as sketched yellow spheres. Yellow shading: section of DNAi complementary to resources. Pink frames: regions of hybridization between DNAi. b: interactions involving species 13.0 including its homodimerization (b3). c: interactions involving species 8.0, including its binding to 13.0 (c3). e: interactions involving species 12.1, including its homodimerization (e3). d and f: DNAi heterodimers interactions suggesting parasitism (d1), and possibly mutualism (d2) and mutual damage (f). The online version of this article includes the following figure supplement(s) for Figure 4: Figure 4-figure supplement 1. Evolution of the intra-species interaction strengths (⟨ΔGpp⟩). Figure 4-figure supplement 2. 2D probability for a consumer, shown at different experimental cycles, to have a given ⟨ω⟩ jointly with a given attachment coordinate to a target strand.