Extrapolating missing antibody-virus measurements across serological studies

Einav, Tal; Cleary, Brian

doi:10.1016/j.cels.2022.06.001

Cited by 7 publications

(21 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To compare our results, we reran our leave-multi-out analysis from Figure 4, simultaneously withholding 124 viruses and predicting their values using an established NNM algorithm from (Einav and Cleary, 2022). The predictions in each table were markedly worse, with the Fonville and Vinh datasets exhibiting σ Actual between 3.4x-5.4x.…”

Section: Resultsmentioning

confidence: 99%

“…As another point of comparison, consider the leave-one-out predictions of the six Vinh viruses using the Fonville datasets. Whereas our algorithm yields tight predictions that span the full range of measured HAI values (Figure S4), NNM led to a nearly flat response with all 25,000 sera incorrectly predicted to be the mean of the measurements [see Figure S8 in (Einav and Cleary, 2022)]. Finally, we emphasize that there is currently no robust method to compute the error of NNM when predicting new virus behavior across datasets, and that error quantification is essential to understand which predictions can be trusted.…”

Section: Resultsmentioning

confidence: 99%

“…Yet much of the heterogeneity of antibody responses collapses when we consider their functional behavior such as binding, inhibition, or neutralization against viruses. Previous work has shown that antibody-virus inhibition data are intrinsically low-dimensional (Lapedes and Farber, 2001), which spurred applications ranging from antigenic maps to the recovery of missing values from partially observed data (Smith et al, 2004;Cai et al, 2010;Ndifon, 2011;Einav and Cleary, 2022). However, these efforts have almost exclusively focused on individual datasets of ferret sera generated under controlled laboratory conditions, circumventing the many obstacles of predicting across heterogeneous human studies.…”

Section: Resultsmentioning

confidence: 99%

“…As in previous studies, all analysis was done on log 10 (HAI titers) because experimental measurements span orders of magnitude, and we did not want to bias the predictions towards the largest values (Einav and Cleary, 2022). Thus, when computing the distribution of errors (histogram in Figures 2-4), each of M, μ , and σ are computed in log 10 .…”

Section: Methodsmentioning

confidence: 99%

“…Whereas past efforts could only predict values for partially-observed viruses within a single dataset ( i . e ., predicting the red squares for the green/blue viruses in Dataset 1 or the blue/gray viruses in Dataset 2) (Cai et al, 2010; Ndifon, 2011; Einav and Cleary, 2022), here we predict data for viruses that do not yet have a single measurement in a dataset.…”

Section: Introductionmentioning

confidence: 96%

See 4 more Smart Citations

Using Interpretable Machine Learning to Massively Increase the Number of Antibody-Virus Interactions Across Studies

Einav

2022

Preprint

Self Cite

View full text Add to dashboard Cite

A central challenge in every field of biology is to use existing measurements to predict the outcomes of future experiments. In this work, we consider the wealth of antibody inhibition data against variants of the influenza virus. Due to this virus's genetic diversity and evolvability, the variants examined in one study will often have little-to-no overlap with other studies, making it difficult to discern common patterns or unify datasets for further analysis. To that end, we develop a computational framework that predicts how an antibody or serum would inhibit any variant from any other study. We use this framework to greatly expand 7 influenza datasets utilizing hemagglutination inhibition, validating our method upon 200,000 existing measurements and predicting more than 2,000,000 new values along with their prediction uncertainties. This data-driven approach does not require any information beyond each virus's name and measurements, and even datasets with as few as 5 viruses can be expanded, making this approach widely applicable. Future influenza studies using hemagglutination inhibition can directly utilize our curated datasets to predict newly measured antibody responses against ≈80 H3N2 influenza viruses from 1968-2011, whereas immunological studies utilizing other viruses or a different assay only need to find a single partially-overlapping dataset to extend their work. In essence, this approach enables a shift in perspective when analyzing data from "what you see is what you get" into "what anyone sees is what everyone gets."

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 96%

See 3 more Smart Citations

Using Interpretable Machine Learning to Massively Increase the Number of Antibody-Virus Interactions Across Studies

Einav

2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

Ultrapotent Broadly Neutralizing Human‐llama Bispecific Antibodies against HIV‐1

Xu,

Zhou,

McKee

et al. 2024

Advanced Science

Self Cite

View full text Add to dashboard Cite

Broadly neutralizing antibodies are proposed as therapeutic and prophylactic agents against HIV‐1, but their potency and breadth are less than optimal. This study describes the immunization of a llama with the prefusion‐stabilized HIV‐1 envelope (Env) trimer, BG505 DS‐SOSIP, and the identification and improvement of potent neutralizing nanobodies recognizing the CD4‐binding site (CD4bs) of vulnerability. Two of the vaccine‐elicited CD4bs‐targeting nanobodies, G36 and R27, when engineered into a triple tandem format with llama IgG2a‐hinge region and human IgG1‐constant region (G36×3‐IgG2a and R27×3‐IgG2a), neutralized 96% of a multiclade 208‐strain panel at geometric mean IC80s of 0.314 and 0.033 µg mL−1, respectively. Cryo‐EM structures of these nanobodies in complex with Env trimer revealed the two nanobodies to neutralize HIV‐1 by mimicking the recognition of the CD4 receptor. To enhance their neutralizing potency and breadth, nanobodies are linked to the light chain of the V2‐apex‐targeting broadly neutralizing antibody, CAP256V2LS. The resultant human‐llama bispecific antibody CAP256L‐R27×3LS exhibited ultrapotent neutralization and breadth exceeding other published HIV‐1 broadly neutralizing antibodies, with pharmacokinetics determined in FcRn‐Fc mice similar to the parent CAP256V2LS. Vaccine‐elicited llama nanobodies, when combined with V2‐apex broadly neutralizing antibodies, may therefore be able to fulfill anti‐HIV‐1 therapeutic and prophylactic clinical goals.

show abstract

Extrapolating missing antibody-virus measurements across serological studies

Cited by 7 publications

References 37 publications

Using Interpretable Machine Learning to Massively Increase the Number of Antibody-Virus Interactions Across Studies

Using Interpretable Machine Learning to Massively Increase the Number of Antibody-Virus Interactions Across Studies

Ultrapotent Broadly Neutralizing Human‐llama Bispecific Antibodies against HIV‐1

Using interpretable machine learning to extend heterogeneous antibody-virus datasets

Contact Info

Product

Resources

About