pith. sign in

arxiv: 2606.21351 · v1 · pith:XJRDK43Unew · submitted 2026-06-19 · 🧬 q-bio.PE

Surveying the adaptive landscapes of 10,000 antibodies

Daniel PGH Wong , Aleksandra M. Walczak , Thierry Mora This is my paper

Pith reviewed 2026-06-26 12:42 UTC · model grok-4.3

classification 🧬 q-bio.PE
keywords antibody affinity maturationconvergent mutationspublic clonotypesadaptive landscapesfitness effectssomatic hypermutationpopulation geneticsB cell lineages
0
0 comments X

The pith

A parameter-free framework using convergent mutations in public clonotypes identifies beneficial antibody mutations and a prevalence-fitness tradeoff across more than 10,000 examples.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a parameter-free population genetic method that examines repeated mutations across B cell lineages with similar starting sequences to map which changes improve antibody function. Applied to data from over 10,000 such lineages in 20 people, the method detects selection that varies by lineage and finds that mutations appearing in more lineages tend to deliver smaller fitness gains. The resulting maps align with mutation patterns observed in antibodies that target SARS-CoV-2 and influenza. The same approach reveals that current antibody language models mostly reflect non-selective patterns unless adjusted to isolate selection signals.

Core claim

By applying a parameter-free population genetic framework to the statistics of convergent affinity maturation in more than 10,000 public clonotypes represented by multiple lineages across 20 healthy individuals, the authors identify widespread signatures of clonotype-dependent selection of individual mutations. They estimate the prevalence and typical fitness effects of mutations across the V gene at the single-site level, uncovering a general tradeoff between prevalence and fitness effect. These inferred landscapes broadly reproduce the statistics of convergent mutation in antibodies specific to SARS-CoV-2 and influenza. The framework also benchmarks predictions from existing antibody langu

What carries the argument

The parameter-free population genetic framework that leverages the statistics of convergent affinity maturation in public clonotypes (B cell lineages sharing similar naive sequences) to identify beneficial mutations.

If this is right

  • Selection acts on mutations in a manner that depends on the specific clonotype rather than uniformly across all antibodies.
  • A tradeoff exists such that mutations observed in more lineages tend to confer smaller fitness improvements.
  • The inferred single-site landscapes reproduce observed convergent mutation frequencies in antibodies against specific pathogens.
  • Antibody language models primarily capture non-selective sequence patterns, but renormalization isolates the selection component.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Antibody design efforts could prioritize mutations predicted to be selected within a target clonotype rather than using averaged landscapes.
  • The approach may generalize to other immune receptors or evolutionary systems that exhibit repeated changes from similar starting points.
  • Collecting data from additional individuals would likely sharpen estimates of how fitness effects vary with mutation prevalence.

Load-bearing premise

The observed statistics of convergent mutations within public clonotypes directly reflect clonotype-dependent positive selection on individual sites, without confounding from shared naive sequence biases or sampling effects across individuals.

What would settle it

Observing that convergent mutation patterns within public clonotypes can be fully accounted for by properties of the naive sequences alone or by sampling variation, or that the inferred beneficial mutations fail to appear at higher rates in actual SARS-CoV-2 and influenza antibody responses.

Figures

Figures reproduced from arXiv: 2606.21351 by Aleksandra M. Walczak, Daniel PGH Wong, Thierry Mora.

Figure 1
Figure 1. Figure 1: TODO: Theoretical sharing via Sonnia a la Maria Ruiz Ortega? (Sharing = expectation from selection-aware VDJ recombination).. Upshot: specific identities of clonotypes do not carry information about shared infection challenges between subjects? 1) Train IGOR model of VDJ on non-productive sequences for each subject (how many sequences necessary?) 2) Train for selection using SONNIA? FOCUS ONLY ON “NAIVE” R… view at source ↗
Figure 2
Figure 2. Figure 2: Sequence-wide and site-level evolutionary parallelism of shared memory clonotypes. a) For focal subject 326651, the number of memory clonotypes shared (same V+J genes, CDR3 ) with every other subject, as a function of lineage size in the focal subject, defined as the number of unique sequences with V gene mutations from germline. Inset: shared clonotypes measured with greater than 10 unique sequences in ea… view at source ↗
Figure 3
Figure 3. Figure 3: Coincidence enrichment of fixed substitutions among lineages of shared clonotypes predicts per-amino acid - per-site substitution rates. c FIG. 3. Landscape of coincident mutations in shared-clonotype lineages. (a) Fraction of lineage pairs sharing the same fixed amino acid substitution at each IMGT-aligned site, summed over mutations at each site, for the 3 most abundant V gene families. Lineage pairs wit… view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4 [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 3
Figure 3. Figure 3: Coincidence enrichment of capture the typical beneficial fixation pro Within the model, we can express the [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5 [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 4
Figure 4. Figure 4: Inferred adaptive landscape predicts statistics of convergent somatic this cumulative prediction is not sensitive to the exact [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: do LMs learn the easiest (?) signal of positive selection in the repertoire a b …YLQMNSLRAEDTAVYYC ----------- WGQGTLVTVSS (Germline) …YYQMNSSRAEDTRVYYC ARGYSYYFQFD WGQGTLVTVSS (Clonotype) V gene CDR3 J gene Measure log-likelihood ( ) based LM score of L89Y in lineage consensus sequence (with all other mutations): score seq. with L89Y — seq. with L89 log ℒ s = log ℒ[ ] log ℒ[ ] score of L89Y Frac. clonotyp… view at source ↗
Figure 5
Figure 5. Figure 5: Testing antibody language models on models (Table I), including general protein language [PITH_FULL_IMAGE:figures/full_fig_p010_5.png] view at source ↗
read the original abstract

Affinity maturation is the Darwinian process by which antibodies improve antigen binding through somatic hypermutation and selection. The adaptive landscape, which defines the set of antibody-specific mutations that improve functional characteristics like antigen binding, has been explored in only a handful of antibodies. Identifying the sites of adaptive mutations in a given antibody sequence, and how these sites vary across the antibody repertoire, can inform the design of therapeutic antibodies. We develop a parameter-free population genetic framework that leverages the statistics of convergent affinity maturation in B cell lineages sharing similar naive sequences, called public clonotypes, to identify beneficial mutations. Applying this framework to more than 10,000 public clonotypes represented by multiple lineages across 20 healthy individuals, we identify widespread signatures of clonotype-dependent selection of individual mutations. We estimate the prevalence and typical fitness effects of mutations across the V gene at the single-site level, uncovering a general tradeoff between prevalence and fitness effect. These inferred landscapes broadly reproduce the statistics of convergent mutation in antibodies specific to SARS-CoV-2 and influenza. Finally, we use our framework to benchmark predictions from existing antibody language models, and show that while these models are dominated by non-selective signatures, a simple renormalization procedure can expose signatures of clonotype-dependent positive selection consistent with our predictions.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper develops a parameter-free population genetic framework that uses statistics of convergent affinity maturation in public clonotypes (B cell lineages sharing similar naive sequences) to infer site-specific beneficial mutations and adaptive landscapes. Applied to >10,000 public clonotypes across 20 healthy individuals, it reports widespread clonotype-dependent selection signatures, a general tradeoff between mutation prevalence and fitness effect, reproduction of convergent mutation statistics in SARS-CoV-2 and influenza antibodies, and a renormalization procedure to extract selection signals from antibody language models.

Significance. If the framework isolates selection without hidden parameters or post-hoc adjustments, the scale of the survey (10k+ clonotypes) and the reproduction of known convergent patterns would represent a substantial advance in mapping antibody adaptive landscapes, with direct relevance to therapeutic design. The explicit parameter-free claim and use for model benchmarking are positive features that could be cited if the central inference holds.

major comments (2)
  1. [Abstract] Abstract: The central inference attributes excess convergence of specific mutations within public clonotypes to clonotype-dependent positive selection. However, no explicit null model or correction is described for mutation-rate heterogeneity or sequence-context biases that could correlate with the naive V-gene similarity used to define clonotypes, which could produce the same patterns under neutrality.
  2. [Abstract] Abstract: The reported 'general tradeoff between prevalence and fitness effect' and the reproduction of convergent mutation statistics rest on unexamined data processing and statistical definitions; without verification that these quantities are extracted directly without fitted parameters or self-referential definitions, the load-bearing claims cannot be assessed.
minor comments (1)
  1. [Abstract] Abstract: The phrase 'broadly reproduce' lacks quantitative metrics (e.g., correlation coefficients or overlap statistics) that would clarify the strength of agreement with SARS-CoV-2 and influenza data.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their detailed comments. We address each major point below with clarifications on the framework's controls and definitions. Our responses focus on the manuscript content without misrepresentation.

read point-by-point responses

  1. Referee: The central inference attributes excess convergence of specific mutations within public clonotypes to clonotype-dependent positive selection. However, no explicit null model or correction is described for mutation-rate heterogeneity or sequence-context biases that could correlate with the naive V-gene similarity used to define clonotypes, which could produce the same patterns under neutrality.

    Authors: The clonotype definition groups lineages by naive V-gene sequence similarity, ensuring comparable mutational contexts within each clonotype. The inference uses the statistic of repeated independent acquisition of the same mutation across lineages of one clonotype, which exceeds the baseline rate observed across clonotypes. This structure controls for context-dependent mutation rates because the starting sequences are matched; global biases uncorrelated with clonotype identity would not produce clonotype-specific convergence patterns. We will revise the abstract and add a methods paragraph explicitly describing this control and why separate null simulations are not required for the parameter-free claim. revision_made = partial revision: partial

  2. Referee: The reported 'general tradeoff between prevalence and fitness effect' and the reproduction of convergent mutation statistics rest on unexamined data processing and statistical definitions; without verification that these quantities are extracted directly without fitted parameters or self-referential definitions, the load-bearing claims cannot be assessed.

    Authors: Prevalence is the direct fraction of clonotypes containing the mutation at least once. The fitness effect is the within-clonotype convergence rate (fraction of lineages acquiring the mutation) normalized by the clonotype's overall mutation count, using raw counts with no fitted parameters. These definitions are independent: prevalence is a global count, while the convergence statistic is local to each clonotype. The SARS-CoV-2 and influenza reproductions apply identical count-based definitions to those datasets. We will add a methods subsection with explicit formulas and verification that no self-reference or fitting occurs. revision_made = yes revision: yes

Circularity Check

0 steps flagged

No circularity: parameter-free inference from empirical convergence counts

full rationale

The framework is explicitly parameter-free and derives site-wise prevalence and fitness-effect estimates directly from observed counts of convergent mutations within public clonotypes defined by naive-sequence similarity. No equations reduce a claimed prediction back to a fitted input by construction, no self-citation chain supplies the core uniqueness or ansatz, and the method does not rename a known empirical pattern under new coordinates. The derivation therefore remains self-contained against the input statistics; any concern about confounding by naive-sequence biases or sampling is a question of external validity rather than internal circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on standard population-genetic assumptions about what convergent mutations indicate; no free parameters are declared, and no new entities are introduced.

axioms (2)
  • domain assumption Convergent mutations across independent lineages sharing similar naive sequences indicate positive selection on those sites.
    Invoked in the description of the framework that leverages statistics of convergent affinity maturation.
  • domain assumption Public clonotypes across healthy individuals provide an unbiased sample for inferring general V-gene selection patterns.
    Used when applying the framework to data from 20 healthy individuals.

pith-pipeline@v0.9.1-grok · 5756 in / 1447 out tokens · 19003 ms · 2026-06-26T12:42:58.528289+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

88 extracted references · 2 canonical work pages

  1. [1]

    MacLennan IC (1994) Germinal centers.Annual review of immunology12:117–139

    1994

  2. [2]

    Annual review of immunology30:429–457

    Victora GD, Nussenzweig MC (2012) Germinal centers. Annual review of immunology30:429–457

    2012

  3. [3]

    Annual review of immunology40:413–442

    Victora GD, Nussenzweig MC (2022) Germinal centers. Annual review of immunology40:413–442

    2022

  4. [4]

    (2025) Replaying germinal center evo- lution on a quantified affinity landscape.Cell

    DeWitt WS, et al. (2025) Replaying germinal center evo- lution on a quantified affinity landscape.Cell

    2025

  5. [5]

    Koenig P, et al. (2015) Deep sequencing-guided design of a high affinity dual specificity antibody to target two angiogenic factors in neovascular age-related macular de- generation.Journal of Biological Chemistry290:21773– 21786

    2015

  6. [6]

    Adams RM, Mora T, Walczak AM, Kinney JB (2016) Measuring the sequence-affinity landscape of antibodies with massively parallel titration curves.Elife5:e23156

    2016

  7. [7]

    Koenig P, et al. (2017) Mutational landscape of anti- body variable domains reveals a switch modulating the interdomain conformational dynamics and antigen bind- ing.Proceedings of the National Academy of Sciences 114:E486–E495

    2017

  8. [8]

    Madan B, et al. (2021) Mutational fitness landscapes reveal genetic and structural improvement pathways for a vaccine-elicited hiv-1 broadly neutralizing anti- body.Proceedings of the National Academy of Sciences 118:e2011653118

    2021

  9. [9]

    (2021) Binding affinity landscapes constrain the evolution of broadly neutralizing anti- influenza antibodies.Elife10:e71393

    Phillips AM, et al. (2021) Binding affinity landscapes constrain the evolution of broadly neutralizing anti- influenza antibodies.Elife10:e71393

    2021

  10. [10]

    (2022) Compensatory epistasis main- tains ace2 affinity in sars-cov-2 omicron ba

    Moulana A, et al. (2022) Compensatory epistasis main- tains ace2 affinity in sars-cov-2 omicron ba. 1.Nature communications13:7011

    2022

  11. [11]

    (2023) The landscape of antibody bind- ing affinity in sars-cov-2 omicron ba

    Moulana A, et al. (2023) The landscape of antibody bind- ing affinity in sars-cov-2 omicron ba. 1 evolution.Elife 12:e83442

    2023

  12. [12]

    Schulz S, Tan TJ, Wu NC, Wang S (2025) Epistatic hotspots organize antibody fitness landscape and boost evolvability.Proceedings of the National Academy of Sci- ences122:e2413884122

    2025

  13. [13]

    (2025) Retrospective sars-cov-2 human antibody development trajectories are largely sparse and permissive.Proceedings of the National Academy of Sci- ences122:e2412787122

    Kirby MB, et al. (2025) Retrospective sars-cov-2 human antibody development trajectories are largely sparse and permissive.Proceedings of the National Academy of Sci- ences122:e2412787122

    2025

  14. [14]

    Nourmohammad A, Otwinowski J, Luksza M, Mora T, Walczak AM (2019) Fierce selection and interference in b-cell repertoire response to chronic hiv-1.Molecular bi- ology and evolution36:2184–2194

    2019

  15. [15]

    Yaari G, Uduman M, Kleinstein SH (2012) Quantifying selection in high-throughput immunoglobulin sequencing data sets.Nucleic acids research40:e134–e134

    2012

  16. [16]

    Yaari G, Benichou JI, Vander Heiden JA, Kleinstein SH, Louzoun Y (2015) The mutation patterns in b-cell im- munoglobulin receptors reflect the influence of selection acting at multiple time-scales.Philosophical Transactions of the Royal Society B: Biological Sciences370

    2015

  17. [17]

    Horns F, Vollmers C, Dekker CL, Quake SR (2019) Signatures of selection in the human antibody reper- toire: Selective sweeps, competing subclones, and neutral drift.Proceedings of the National Academy of Sciences 116:1261–1266

    2019

  18. [18]

    (2021) Human b cell lineages associated with germinal centers following influenza vaccination are measurably evolving.Elife10:e70873

    Hoehn KB, et al. (2021) Human b cell lineages associated with germinal centers following influenza vaccination are measurably evolving.Elife10:e70873

    2021

  19. [19]

    Ralph DK, Matsen IV FA (2020) Using b cell receptor lineage structures to predict affinity.PLOS Computa- tional Biology16:e1008391

    2020

  20. [20]

    (2022) Memory persistence and differ- entiation into antibody-secreting cells accompanied by positive selection in longitudinal bcr repertoires.Elife 11:e79254

    Mikelov A, et al. (2022) Memory persistence and differ- entiation into antibody-secreting cells accompanied by positive selection in longitudinal bcr repertoires.Elife 11:e79254

    2022

  21. [21]

    Ruffolo JA, Gray JJ, Sulam J (2021) Decipher- ing antibody affinity maturation with language mod- els and weakly supervised learning.arXiv preprint arXiv:2112.07782

    work page arXiv 2021

  22. [22]

    Bioinformatics Advances2:vbac046

    Olsen TH, Moal IH, Deane CM (2022) Ablang: an anti- body language model for completing antibody sequences. Bioinformatics Advances2:vbac046

    2022

  23. [23]

    Shuai RW, Ruffolo JA, Gray JJ (2023) Iglm: Infilling language modeling for antibody sequence design.Cell systems14:979–989

    2023

  24. [24]

    (2024) Large scale paired anti- body language models.PLOS Computational Biology 20:e1012646

    Kenlay H, et al. (2024) Large scale paired anti- body language models.PLOS Computational Biology 20:e1012646

    2024

  25. [25]

    Burbach SM, Briney B (2024) Improving antibody lan- guage models with native pairing.Patterns5

    2024

  26. [26]

    Burbach SM, Briney B (2025) A curriculum learning approach to training antibody language models.PLOS Computational Biology21:e1013473

    2025

  27. [27]

    Olsen TH, Moal IH, Deane CM (2024) Addressing the an- tibody germline bias and its effect on language models for 15 improved antibody design.Bioinformatics40:btae618

    2024

  28. [28]

    Ng K, Briney B (2025) Focused learning by anti- body language models using preferential masking of non- templated regions.Patterns6

    2025

  29. [29]

    (2025) A sitewise model of natu- ral selection on individual antibodies via a transformer– encoder.Molecular Biology and Evolution42:msaf186

    Matsen IV FA, et al. (2025) A sitewise model of natu- ral selection on individual antibodies via a transformer– encoder.Molecular Biology and Evolution42:msaf186

    2025

  30. [30]

    (2026) Separating selection from mutation in antibody language models.Elife 15:RP109644

    Matsen IV FA, et al. (2026) Separating selection from mutation in antibody language models.Elife 15:RP109644

    2026

  31. [31]

    (2018) Multi-donor longitudinal antibody repertoire sequencing reveals the existence of public an- tibody clonotypes in hiv-1 infection.Cell host & microbe 23:845–854

    Setliff I, et al. (2018) Multi-donor longitudinal antibody repertoire sequencing reveals the existence of public an- tibody clonotypes in hiv-1 infection.Cell host & microbe 23:845–854

    2018

  32. [32]

    (2022) Sequence and functional char- acterization of a public hiv-specific antibody clonotype

    Murji AA, et al. (2022) Sequence and functional char- acterization of a public hiv-specific antibody clonotype. Iscience25

    2022

  33. [33]

    (2023) Convergent antibody responses are associated with broad neutralization of hepatitis c virus.Frontiers in immunology14:1135841

    Skinner NE, et al. (2023) Convergent antibody responses are associated with broad neutralization of hepatitis c virus.Frontiers in immunology14:1135841

    2023

  34. [34]

    (2019) Polyclonal and convergent an- tibody response to ebola virus vaccine rvsv-zebov.Nature medicine25:1589–1600

    Ehrhardt SA, et al. (2019) Polyclonal and convergent an- tibody response to ebola virus vaccine rvsv-zebov.Nature medicine25:1589–1600

    2019

  35. [35]

    (2021) Convergent antibody evolution and clonotype expansion following influenza virus vacci- nation.PLoS One16:e0247253

    Forgacs D, et al. (2021) Convergent antibody evolution and clonotype expansion following influenza virus vacci- nation.PLoS One16:e0247253

    2021

  36. [36]

    (2024) An explainable language model for antibody specificity prediction using curated influenza hemagglutinin antibodies.Immunity57:2453–2465

    Wang Y, et al. (2024) An explainable language model for antibody specificity prediction using curated influenza hemagglutinin antibodies.Immunity57:2453–2465

    2024

  37. [37]

    (2020) Convergent antibody re- sponses to sars-cov-2 in convalescent individuals.Nature 584:437–442

    Robbiani DF, et al. (2020) Convergent antibody re- sponses to sars-cov-2 in convalescent individuals.Nature 584:437–442

    2020

  38. [38]

    (2020) Deep sequencing of b cell re- ceptor repertoires from covid-19 patients reveals strong convergent immune signatures.Frontiers in immunology 11:605170

    Galson JD, et al. (2020) Deep sequencing of b cell re- ceptor repertoires from covid-19 patients reveals strong convergent immune signatures.Frontiers in immunology 11:605170

    2020

  39. [39]

    (2021) Convergent antibody responses to the sars-cov-2 spike protein in convalescent and vacci- nated individuals.Cell reports36

    Chen EC, et al. (2021) Convergent antibody responses to the sars-cov-2 spike protein in convalescent and vacci- nated individuals.Cell reports36

    2021

  40. [40]

    (2021) Sequence signatures of two public antibody clonotypes that bind sars-cov-2 receptor bind- ing domain.Nature Communications12:3815

    Tan TJ, et al. (2021) Sequence signatures of two public antibody clonotypes that bind sars-cov-2 receptor bind- ing domain.Nature Communications12:3815

    2021

  41. [41]

    (2022) A large-scale systematic survey reveals recurring molecular features of public antibody responses to sars-cov-2.Immunity55:1105–1117

    Wang Y, et al. (2022) A large-scale systematic survey reveals recurring molecular features of public antibody responses to sars-cov-2.Immunity55:1105–1117

    2022

  42. [42]

    (2022) The prominent role of a cdr1 somatic hypermutation for convergent ighv3-53/3-66 antibodies in binding to sars-cov-2.Emerging Microbes & Infections 11:1186–1190

    Tian X, et al. (2022) The prominent role of a cdr1 somatic hypermutation for convergent ighv3-53/3-66 antibodies in binding to sars-cov-2.Emerging Microbes & Infections 11:1186–1190

    2022

  43. [43]

    (2025) Clonotype-enriched somatic hyper- mutations drive affinity maturation of a public human an- tibody targeting an occluded sarbecovirus epitope.Cell reports44

    Rao VN, et al. (2025) Clonotype-enriched somatic hyper- mutations drive affinity maturation of a public human an- tibody targeting an occluded sarbecovirus epitope.Cell reports44

    2025

  44. [44]

    (2025) Ai identifies broadly neutralizing an- tibodies from an ighv1-69 public antibody class exerting continued selection over sars-cov-2.bioRxivpp 2025–12

    Niu C, et al. (2025) Ai identifies broadly neutralizing an- tibodies from an ighv1-69 public antibody class exerting continued selection over sars-cov-2.bioRxivpp 2025–12

    2025

  45. [45]

    Briney B, Inderbitzin A, Joyce C, Burton DR (2019) Commonality despite exceptional diversity in the baseline human antibody repertoire.Nature566:393–397

    2019

  46. [46]

    (2022) Clonal structure, stability and dynamics of human memory b cells and circulating plas- mablasts.Nature immunology23:1076–1085

    Phad GE, et al. (2022) Clonal structure, stability and dynamics of human memory b cells and circulating plas- mablasts.Nature immunology23:1076–1085

    2022

  47. [47]

    Cvijovi´ c I, Swift M, Quake SR (2025) Long-term b cell memory emerges at uniform relative rates in the human immune response.Proceedings of the National Academy of Sciences122:e2406474122

    2025

  48. [48]

    Spisak N, Ath` enes G, Dupic T, Mora T, Walczak AM (2024) Combining mutation and recombination statis- tics to infer clonal families in antibody repertoires.Elife 13:e86181

    2024

  49. [49]

    Ruiz Ortega M, Spisak N, Mora T, Walczak AM (2023) Modeling and predicting the overlap of b-and t-cell re- ceptor repertoires in healthy and sars-cov-2 infected in- dividuals.PLoS Genetics19:e1010652

    2023

  50. [50]

    (2015) Inferring processes underlying b-cell repertoire diversity.Philosophical Transactions of the Royal Society B: Biological Sciences370

    Elhanati Y, et al. (2015) Inferring processes underlying b-cell repertoire diversity.Philosophical Transactions of the Royal Society B: Biological Sciences370

    2015

  51. [51]

    Marcou Q, Mora T, Walczak AM (2018) High- throughput immune repertoire analysis with igor.Nature communications9:561

    2018

  52. [52]

    Sethna Z, Elhanati Y, Callan Jr CG, Walczak AM, Mora T (2019) Olga: fast computation of generation proba- bilities of b-and t-cell receptor amino acid sequences and motifs.Bioinformatics35:2974–2981

    2019

  53. [53]

    Isacchini G, Walczak AM, Mora T, Nourmohammad A (2021) Deep generative selection models of t and b cell receptor repertoires with sonnia.Proceedings of the Na- tional Academy of Sciences118:e2023141118

    2021

  54. [54]

    Desponds J, Mora T, Walczak AM (2016) Fluctuating fit- ness shapes the clone-size distribution of immune reper- toires.Proceedings of the National Academy of Sciences 113:274–279

    2016

  55. [55]

    arXiv preprint arXiv:2510.02812

    Mazzolini A, Walczak AM, Mora T (2025) Dynamics of memory b cells and plasmablasts in healthy individuals. arXiv preprint arXiv:2510.02812

    work page arXiv 2025

  56. [56]

    (1989) Conformations of immunoglob- ulin hypervariable regions.Nature342:877–883

    Chothia C, et al. (1989) Conformations of immunoglob- ulin hypervariable regions.Nature342:877–883

    1989

  57. [57]

    (2015) Quantifying evolutionary constraints on b-cell affinity maturation.Philosophical Transactions of the Royal Society B: Biological Sciences 370

    McCoy CO, et al. (2015) Quantifying evolutionary constraints on b-cell affinity maturation.Philosophical Transactions of the Royal Society B: Biological Sciences 370

    2015

  58. [58]

    (2022) Broadly neutralizing anti- bodies target a haemagglutinin anchor epitope.Nature 602:314–320

    Guthmiller JJ, et al. (2022) Broadly neutralizing anti- bodies target a haemagglutinin anchor epitope.Nature 602:314–320

    2022

  59. [59]

    Raju N, et al. (2024) Multiplexed antibody sequencing and profiling of the human hemagglutinin-specific mem- ory b cell response following influenza vaccination.The Journal of Immunology213:1605–1619

    2024

  60. [60]

    (2026) B cell imprinting in children impairs antibodies to the haemagglutinin stalk.Naturepp 1–10

    Sun J, et al. (2026) B cell imprinting in children impairs antibodies to the haemagglutinin stalk.Naturepp 1–10

    2026

  61. [61]

    (2020) Structural basis of a shared anti- body response to sars-cov-2.Science369:1119–1123

    Yuan M, et al. (2020) Structural basis of a shared anti- body response to sars-cov-2.Science369:1119–1123

    2020

  62. [62]

    Chungyoun M, Gray J (2025) Fitness landscape for an- tibodies 2: Benchmarking reveals that protein ai models cannot yet consistently predict developability properties. bioRxiv

    2025

  63. [63]

    Nijkamp E, Ruffolo JA, Weinstein EN, Naik N, Madani A (2023) Progen2: exploring the boundaries of protein language models.Cell systems14:968–978

    2023

  64. [64]

    (2023) Evolutionary-scale prediction of atomic-level protein structure with a language model

    Lin Z, et al. (2023) Evolutionary-scale prediction of atomic-level protein structure with a language model. Science379:1123–1130

    2023

  65. [65]

    (2026) Scaling unlocks broader generation and deeper functional understanding of pro- teins.Advances in Neural Information Processing Sys- tems38:46109–46145

    Bhatnagar A, et al. (2026) Scaling unlocks broader generation and deeper functional understanding of pro- teins.Advances in Neural Information Processing Sys- tems38:46109–46145

    2026

  66. [66]

    Protein Science31:141–146

    Olsen TH, Boyles F, Deane CM (2022) Observed anti- body space: A diverse database of cleaned, annotated, 16 and translated unpaired and paired antibody sequences. Protein Science31:141–146

    2022

  67. [67]

    (2021) Language models enable zero-shot prediction of the effects of mutations on protein func- tion.Advances in neural information processing systems 34:29287–29303

    Meier J, et al. (2021) Language models enable zero-shot prediction of the effects of mutations on protein func- tion.Advances in neural information processing systems 34:29287–29303

    2021

  68. [68]

    (2022)Tranception: protein fitness predic- tion with autoregressive transformers and inference-time retrieval(PMLR), pp 16990–17017

    Notin P, et al. (2022)Tranception: protein fitness predic- tion with autoregressive transformers and inference-time retrieval(PMLR), pp 16990–17017

    2022

  69. [69]

    (2021) Optimization of therapeutic an- tibodies by predicting antigen specificity from antibody sequence via deep learning.Nature biomedical engineer- ing5:600–612

    Mason DM, et al. (2021) Optimization of therapeutic an- tibodies by predicting antigen specificity from antibody sequence via deep learning.Nature biomedical engineer- ing5:600–612

    2021

  70. [70]

    Pugh CW, Nu˜ nez-Valencia PG, Dias M, Frazer J (2026) From likelihood to fitness: Improving variant effect pre- diction in protein and genome language models.Advances in Neural Information Processing Systems38:130835– 130866

    2026

  71. [71]

    (2026) Conditionally site-independent neu- ral evolution of antibody sequences.ArXivpp arXiv– 2602

    Lu SZ, et al. (2026) Conditionally site-independent neu- ral evolution of antibody sequences.ArXivpp arXiv– 2602

    2026

  72. [72]

    Molari M, Eyer K, Baudry J, Cocco S, Monasson R (2020) Quantitative modeling of the effect of antigen dosage on b-cell affinity distributions in maturating ger- minal centers.Elife9:e55678

    2020

  73. [73]

    (2026) Inference of germinal center evo- lutionary dynamics via simulation-based deep learning

    Ralph DK, et al. (2026) Inference of germinal center evo- lutionary dynamics via simulation-based deep learning. eLife14:RP108880

    2026

  74. [74]

    Nucleic acids research41:W34–W40

    Ye J, Ma N, Madden TL, Ostell JM (2013) Igblast: an immunoglobulin variable domain sequence analysis tool. Nucleic acids research41:W34–W40

    2013

  75. [75]

    Gadala-Maria D, Yaari G, Uduman M, Kleinstein SH (2015) Automated analysis of high-throughput b-cell se- quencing data reveals a high frequency of novel im- munoglobulin v gene segment alleles.Proceedings of the National Academy of Sciences112:E862–E870

    2015

  76. [76]

    Raybould MI, Kovaltsuk A, Marks C, Deane CM (2021) Cov-abdab: the coronavirus antibody database.Bioin- formatics37:734–735

    2021

  77. [77]

    (2015) Mixcr: software for compre- hensive adaptive immunity profiling.Nature methods 12:380–381

    Bolotin DA, et al. (2015) Mixcr: software for compre- hensive adaptive immunity profiling.Nature methods 12:380–381

    2015

  78. [78]

    (2013) Models of somatic hypermutation targeting and substitution based on synonymous muta- tions from high-throughput immunoglobulin sequencing data.Frontiers in immunology4:358

    Yaari G, et al. (2013) Models of somatic hypermutation targeting and substitution based on synonymous muta- tions from high-throughput immunoglobulin sequencing data.Frontiers in immunology4:358

    2013

  79. [79]

    Spisak N, Walczak AM, Mora T (2020) Learning the heterogeneous hypermutation landscape of immunoglob- ulins from high-throughput repertoire data.Nucleic acids research48:10702–10712

    2020

  80. [80]

    mikelov-et-al-2021

    Good BH, Rouzine IM, Balick DJ, Hallatschek O, Desai MM (2012) Distribution of fixed beneficial mutations and the rate of adaptation in asexual populations.Proceed- ings of the National Academy of Sciences109:4950–4955. 17 Appendix A: Deep repertoire sequencing datasets The B cell repertoire datasets analyzed in this study fulfilled three criteria. Each s...

    2012

Showing first 80 references.