Department of Statistics, Stanford University, Stanford, California, United States of America *Authors contributed equally to this work Correspondence should be addressed to teinav@fredhutch.org Abstract 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 seven influenza datasets utilizing hemagglutination inhibition, validating our method upon 200,000 existing measurements and predicting 2,000,000 new values uncertainties. With these new values, we quantify the transferability between seven vaccination and infection studies in humans and ferrets, show that the serum potency is negatively correlated with breadth, and present a tool for pandemic preparedness. 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 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." Introduction Our understanding of how antibody-mediated immunity drives viral evolution and escape relies upon painstaking measurements of antibody binding, inhibition, or neutralization against variants of concern (Petrova and Russell, 2017). Every interaction is unique because: (1) the antibody response (serum) changes even in the absence of viral exposure and (2) for rapidly evolving viruses such as influenza, the specific variants examined in one study will often have little-to-no overlap with other studies (Figure 1).

Temporal knowledge graph (TKG) reasoning is a crucial task that has gained increasing research interest in recent years. Most existing methods focus on reasoning at past timestamps to complete the missing facts, and there are only a few works of reasoning on known TKGs to forecast future facts. Compared with the completion task, the forecasting task is more difficult that faces two main challenges: (1) how to effectively model the time information to handle future timestamps? (2) how to make inductive inference to handle previously unseen entities that emerge over time? To address these challenges, we propose the first reinforcement learning method for forecasting. Specifically, the agent travels on historical knowledge graph snapshots to search for the answer. Our method defines a relative time encoding function to capture the timespan information, and we design a novel time-shaped reward based on Dirichlet distribution to guide the model learning. Furthermore, we propose a novel representation method for unseen entities to improve the inductive inference ability of the model. We evaluate our method for this link prediction task at future timestamps. Extensive experiments on four benchmark datasets demonstrate substantial performance improvement meanwhile with higher explainability, less calculation, and fewer parameters when compared with existing state-of-the-art methods.

Immune memory against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) helps to determine protection against reinfection, disease risk, and vaccine efficacy. Using 188 human cases across the range of severity of COVID-19, Dan et al. analyzed cross-sectional data describing the dynamics of SARS-CoV-2 memory B cells, CD8+ T cells, and CD4+ T cells for more than 6 months after infection. The authors found a high degree of heterogeneity in the magnitude of adaptive immune responses that persisted into the immune memory phase to the virus. However, immune memory in three immunological compartments remained measurable in greater than 90% of subjects for more than 5 months after infection. Despite the heterogeneity of immune responses, these results show that durable immunity against secondary COVID-19 disease is a possibility for most individuals. Science , this issue p. [eabf4063][1] ### INTRODUCTION Immunological memory is the basis for durable protective immunity after infections or vaccinations. Duration of immunological memory after severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and COVID-19 is unclear. Immunological memory can consist of memory B cells, antibodies, memory CD4+ T cells, and/or memory CD8+ T cells. Knowledge of the kinetics and interrelationships among those four types of memory in humans is limited. Understanding immune memory to SARS-CoV-2 has implications for understanding protective immunity against COVID-19 and assessing the likely future course of the COVID-19 pandemic. ### RATIONALE Assessing virus-specific immune memory over at least a 6-month period is likely necessary to ascertain the durability of immune memory to SARS-CoV-2. Given the evidence that antibodies, CD4+ T cells, and CD8+ T cells can all participate in protective immunity to SARS-CoV-2, we measured antigen-specific antibodies, memory B cells, CD4+ T cells, and CD8+ T cells in the blood from subjects who recovered from COVID-19, up to 8 months after infection. ### RESULTS The study involved 254 samples from 188 COVID-19 cases, including 43 samples at 6 to 8 months after infection. Fifty-one subjects in the study provided longitudinal blood samples, allowing for both cross-sectional and longitudinal analyses of SARS-CoV-2–specific immune memory. Antibodies against SARS-CoV-2 spike and receptor binding domain (RBD) declined moderately over 8 months, comparable to several other reports. Memory B cells against SARS-CoV-2 spike actually increased between 1 month and 8 months after infection. Memory CD8+ T cells and memory CD4+ T cells declined with an initial half-life of 3 to 5 months. This is the largest antigen-specific study to date of the four major types of immune memory for any viral infection. Among the antibody responses, spike immunoglobulin G (IgG), RBD IgG, and neutralizing antibody titers exhibited similar kinetics. Spike IgA was still present in the large majority of subjects at 6 to 8 months after infection. Among the memory B cell responses, IgG was the dominant isotype, with a minor population of IgA memory B cells. IgM memory B cells appeared to be short-lived. CD8+ T cell and CD4+ T cell memory was measured for all SARS-CoV-2 proteins. Although ~70% of individuals possessed detectable CD8+ T cell memory at 1 month after infection, that proportion declined to ~50% by 6 to 8 months after infection. For CD4+ T cell memory, 93% of subjects had detectable SARS-CoV-2 memory at 1 month after infection, and the proportion of subjects positive for CD4+ T cells (92%) remained high at 6 to 8 months after infection. SARS-CoV-2 spike-specific memory CD4+ T cells with the specialized capacity to help B cells [T follicular helper (TFH) cells] were also maintained. The different types of immune memory each had distinct kinetics, resulting in complex interrelationships between the abundance of T cell, B cell, and antibody immune memory over time. Additionally, substantially heterogeneity in memory to SARS-CoV-2 was observed. ### CONCLUSION Substantial immune memory is generated after COVID-19, involving all four major types of immune memory. About 95% of subjects retained immune memory at ~6 months after infection. Circulating antibody titers were not predictive of T cell memory. Thus, simple serological tests for SARS-CoV-2 antibodies do not reflect the richness and durability of immune memory to SARS-CoV-2. This work expands our understanding of immune memory in humans. These results have implications for protective immunity against SARS-CoV-2 and recurrent COVID-19. ![Figure][2] Immunological memory consists of antibodies, memory B cells, memory CD8+ T cells, and memory CD4+ T cells. This study examined all of the types of virus-specific immune memory against SARS-CoV-2 in COVID-19 subjects. Robust immune memory was observed in most individuals. Understanding immune memory to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is critical for improving diagnostics and vaccines and for assessing the likely future course of the COVID-19 pandemic. We analyzed multiple compartments of circulating immune memory to SARS-CoV-2 in 254 samples from 188 COVID-19 cases, including 43 samples at ≥6 months after infection. Immunoglobulin G (IgG) to the spike protein was relatively stable over 6+ months. Spike-specific memory B cells were more abundant at 6 months than at 1 month after symptom onset. SARS-CoV-2–specific CD4+ T cells and CD8+ T cells declined with a half-life of 3 to 5 months. By studying antibody, memory B cell, CD4+ T cell, and CD8+ T cell memory to SARS-CoV-2 in an integrated manner, we observed that each component of SARS-CoV-2 immune memory exhibited distinct kinetics. [1]: /lookup/doi/10.1126/science.abf4063 [2]: pending:yes

The infection dynamics of Zika virus (ZIKV) are difficult to characterize. Many ZIKV infections are asymptomatic, and the clinical presentation of ZIKV is nonspecific. Rodriguez-Barraquer et al. took advantage of a long-term health study under way in Salvador, Brazil, the epicenter of the recent outbreak in the Americas. They used multiple serological assays, from before and after the emergence of ZIKV in October 2015, to distinguish ZIKV immune responses from those against Dengue virus (DENV). About 73% of the population was attacked by ZIKV.