Milena Pavlovic

• Victoriano, Margarida; Pavlovic, Milena; Sandve, Geir Kjetil Ferkingstad; Oliveira, Hélder P.; Rocha, Artur & Greiff, Victor (2026). From virtual experiments to biomedical insight with synthetic data. Nature Machine Intelligence. 8, s. 866–879. doi: 10.1038/s42256-026-01244-6. Fulltekst i vitenarkiv
• Chernigovskaia, Maria; Pavlović, Milena; Kanduri, Chakravarthi; Gielis, Sofie; Robert, Philippe & Scheffer, Lonneke [Vis alle 12 forfattere av denne artikkelen] (2025). Simulation of adaptive immune receptors and repertoires with complex immune information to guide the development and benchmarking of AIRR machine learning. Nucleic Acids Research (NAR). ISSN 0305-1048. 53(3). doi: 10.1093/nar/gkaf025. Fulltekst i vitenarkiv Vis sammendrag

  Machine learning (ML) has shown great potential in the adaptive immune receptor repertoire (AIRR) field. However, there is a lack of large-scale ground-truth experimental AIRR data suitable for AIRR-ML-based disease diagnostics and therapeutics discovery. Simulated ground-truth AIRR data are required to complement the development and benchmarking of robust and interpretable AIRR-ML methods where experimental data is currently inaccessible or insufficient. The challenge for simulated data to be useful is incorporating key features observed in experimental repertoires. These features, such as antigen or disease-associated immune information, cause AIRR-ML problems to be challenging. Here, we introduce LIgO, a software suite, which simulates AIRR data for the development and benchmarking of AIRR-ML methods. LIgO incorporates different types of immune information both on the receptor and the repertoire level and preserves native-like generation probability distribution. Additionally, LIgO assists users in determining the computational feasibility of their simulations. We show two examples where LIgO supports the development and validation of AIRR-ML methods: (i) how individuals carrying out-of-distribution immune information impacts receptor-level prediction performance and (ii) how immune information co-occurring in the same AIRs impacts the performance of conventional receptor-level encoding and repertoire-level classification approaches. LIgO guides the advancement and assessment of interpretable AIRR-ML methods.

• Hajj, Ghadi al; Hubin, Aliaksandr; Kanduri, Chakravarthi; Pavlovic, Milena; Rand, Knut Dagestad & Widrich, Michael [Vis alle 11 forfattere av denne artikkelen] (2024). Incorporating probabilistic domain knowledge into deep multiple instance learning. Proceedings of Machine Learning Research (PMLR). 235, s. 17279–17297. Fulltekst i vitenarkiv Vis sammendrag

  Deep learning methods, including deep multiple instance learning methods, have been criticized for their limited ability to incorporate domain knowledge. A reason that knowledge incorporation is challenging in deep learning is that the models usually lack a mapping between their model components and the entities of the domain, making it a non-trivial task to incorporate probabilistic prior information. In this work, we show that such a mapping between domain entities and model components can be defined for a multiple instance learning setting and propose a framework DeeMILIP that encompasses multiple strategies to exploit this mapping for prior knowledge incorporation. We motivate and formalize these strategies from a probabilistic perspective. Experiments on an immune-based diagnostics case show that our proposed strategies allow to learn generalizable models even in settings with weak signals, limited dataset size, and limited compute.

• Pavlović, Milena; Hajj, Ghadi al; Kanduri, Chakravarthi; Pensar, Johan; Wood, Mollie Elizabeth & Sollid, Ludvig Magne [Vis alle 8 forfattere av denne artikkelen] (2024). Improving generalization of machine learning-identified biomarkers using causal modelling with examples from immune receptor diagnostics. Nature Machine Intelligence. 6(1), s. 15–24. doi: 10.1038/s42256-023-00781-8. Fulltekst i vitenarkiv
• Scheffer, Lonneke; Reber, Eric Emanuel; Mehta, Brij Bhushan; Pavlović, Milena; Chernigovskaia, Maria & Richardson, Eve [Vis alle 11 forfattere av denne artikkelen] (2024). Predictability of antigen binding based on short motifs in the antibody CDRH3. Briefings in Bioinformatics. ISSN 1467-5463. 25(6), s. 1–10. doi: 10.1093/bib/bbae537. Fulltekst i vitenarkiv Vis sammendrag

  Adaptive immune receptors, such as antibodies and T-cell receptors, recognize foreign threats with exquisite specificity. A major challenge in adaptive immunology is discovering the rules governing immune receptor–antigen binding in order to predict the antigen binding status of previously unseen immune receptors. Many studies assume that the antigen binding status of an immune receptor may be determined by the presence of a short motif in the complementarity determining region 3 (CDR3), disregarding other amino acids. To test this assumption, we present a method to discover short motifs which show high precision in predicting antigen binding and generalize well to unseen simulated and experimental data. Our analysis of a mutagenesis-based antibody dataset reveals 11 336 position-specific, mostly gapped motifs of 3–5 amino acids that retain high precision on independently generated experimental data. Using a subset of only 178 motifs, a simple classifier was made that on the independently generated dataset outperformed a deep learning model proposed specifically for such datasets. In conclusion, our findings support the notion that for some antibodies, antigen binding may be largely determined by a short CDR3 motif. As more experimental data emerge, our methodology could serve as a foundation for in-depth investigations into antigen binding signals.

• Kanduri, Chakravarthi; Scheffer, Lonneke; Pavlović, Milena; Rand, Knut Dagestad; Chernigovskaia, Maria & Pirvandy, Oz [Vis alle 9 forfattere av denne artikkelen] (2023). simAIRR: simulation of adaptive immune repertoires with realistic receptor sequence sharing for benchmarking of immune state prediction methods. GigaScience. 12, s. 1–16. doi: 10.1093/gigascience/giad074. Fulltekst i vitenarkiv Vis sammendrag

  Abstract Background Machine learning (ML) has gained significant attention for classifying immune states in adaptive immune receptor repertoires (AIRRs) to support the advancement of immunodiagnostics and therapeutics. Simulated data are crucial for the rigorous benchmarking of AIRR-ML methods. Existing approaches to generating synthetic benchmarking datasets result in the generation of naive repertoires missing the key feature of many shared receptor sequences (selected for common antigens) found in antigen-experienced repertoires. Results We demonstrate that a common approach to generating simulated AIRR benchmark datasets can introduce biases, which may be exploited for undesired shortcut learning by certain ML methods. To mitigate undesirable access to true signals in simulated AIRR datasets, we devised a simulation strategy (simAIRR) that constructs antigen-experienced-like repertoires with a realistic overlap of receptor sequences. simAIRR can be used for constructing AIRR-level benchmarks based on a range of assumptions (or experimental data sources) for what constitutes receptor-level immune signals. This includes the possibility of making or not making any prior assumptions regarding the similarity or commonality of immune state–associated sequences that will be used as true signals. We demonstrate the real-world realism of our proposed simulation approach by showing that basic ML strategies perform similarly on simAIRR-generated and real-world experimental AIRR datasets. Conclusions This study sheds light on the potential shortcut learning opportunities for ML methods that can arise with the state-of-the-art way of simulating AIRR datasets. simAIRR is available as a Python package: https://github.com/KanduriC/simAIRR.

• Shakibfar, Saeed; Nyberg, Fredrik; Li, Huiqi; Zhao, Jing; Nordeng, Hedvig Marie Egeland & Sandve, Geir Kjetil Ferkingstad [Vis alle 10 forfattere av denne artikkelen] (2023). Artificial intelligence-driven prediction of COVID-19-related hospitalization and death: a systematic review. Frontiers in Public Health. 11. doi: 10.3389/fpubh.2023.1183725. Fulltekst i vitenarkiv Vis sammendrag

  Aim To perform a systematic review on the use of Artificial Intelligence (AI) techniques for predicting COVID-19 hospitalization and mortality using primary and secondary data sources. Study eligibility criteria Cohort, clinical trials, meta-analyses, and observational studies investigating COVID-19 hospitalization or mortality using artificial intelligence techniques were eligible. Articles without a full text available in the English language were excluded. Data sources Articles recorded in Ovid MEDLINE from 01/01/2019 to 22/08/2022 were screened. Data extraction We extracted information on data sources, AI models, and epidemiological aspects of retrieved studies. Bias assessment A bias assessment of AI models was done using PROBAST. Participants Patients tested positive for COVID-19. Results We included 39 studies related to AI-based prediction of hospitalization and death related to COVID-19. The articles were published in the period 2019-2022, and mostly used Random Forest as the model with the best performance. AI models were trained using cohorts of individuals sampled from populations of European and non-European countries, mostly with cohort sample size <5,000. Data collection generally included information on demographics, clinical records, laboratory results, and pharmacological treatments (i.e., high-dimensional datasets). In most studies, the models were internally validated with cross-validation, but the majority of studies lacked external validation and calibration. Covariates were not prioritized using ensemble approaches in most of the studies, however, models still showed moderately good performances with Area under the Receiver operating characteristic Curve (AUC) values >0.7. According to the assessment with PROBAST, all models had a high risk of bias and/or concern regarding applicability. Conclusions A broad range of AI techniques have been used to predict COVID-19 hospitalization and mortality. The studies reported good prediction performance of AI models, however, high risk of bias and/or concern regarding applicability were detected.

• Shakibfar, Saeed; Zhao, Jing; Li, Huiqi; Nordeng, Hedvig Marie Egeland; Lupattelli, Angela & Pavlović, Milena [Vis alle 12 forfattere av denne artikkelen] (2023). Machine learning-driven development of a disease risk score for COVID-19 hospitalization and mortality: a Swedish and Norwegian register-based study. Frontiers in Public Health. 11. doi: 10.3389/fpubh.2023.1258840. Fulltekst i vitenarkiv Vis sammendrag

  Aims To develop a disease risk score for COVID-19-related hospitalization and mortality in Sweden and externally validate it in Norway. Method We employed linked data from the national health registries of Sweden and Norway to conduct our study. We focused on individuals in Sweden with confirmed SARS-CoV-2 infection through RT-PCR testing up to August 2022 as our study cohort. Within this group, we identified hospitalized cases as those who were admitted to the hospital within 14 days of testing positive for SARS-CoV-2 and matched them with five controls from the same cohort who were not hospitalized due to SARS-CoV-2. Additionally, we identified individuals who died within 30 days after being hospitalized for COVID-19. To develop our disease risk scores, we considered various factors, including demographics, infectious, somatic, and mental health conditions, recorded diagnoses, and pharmacological treatments. We also conducted age-specific analyses and assessed model performance through 5-fold cross-validation. Finally, we performed external validation using data from the Norwegian population with COVID-19 up to December 2021. Results During the study period, a total of 124,560 individuals in Sweden were hospitalized, and 15,877 individuals died within 30 days following COVID-19 hospitalization. Disease risk scores for both hospitalization and mortality demonstrated predictive capabilities with ROC-AUC values of 0.70 and 0.72, respectively, across the entire study period. Notably, these scores exhibited a positive correlation with the likelihood of hospitalization or death. In the external validation using data from the Norwegian COVID-19 population (consisting of 53,744 individuals), the disease risk score predicted hospitalization with an AUC of 0.47 and death with an AUC of 0.74. Conclusion The disease risk score showed moderately good performance to predict COVID-19-related mortality but performed poorly in predicting hospitalization when externally validated.

• Robert, Philippe Paul Auguste; Akbar, Rahmad; Frank, Robert; Pavlović, Milena; Widrich, Michael & Snapkov, Igor [Vis alle 25 forfattere av denne artikkelen] (2022). Unconstrained generation of synthetic antibody–antigen structures to guide machine learning methodology for antibody specificity prediction. Nature Computational Science. 2(12), s. 845–865. doi: 10.1038/s43588-022-00372-4. Fulltekst i vitenarkiv
• Kanduri, Chakravarthi; Pavlović, Milena; Scheffer, Lonneke; Motwani, Keshav; Chernigovskaia, Maria & Greiff, Victor [Vis alle 7 forfattere av denne artikkelen] (2022). Profiling the baseline performance and limits of machine learning models for adaptive immune receptor repertoire classification. GigaScience. 11(05). doi: 10.1093/gigascience/giac046. Fulltekst i vitenarkiv Vis sammendrag

  Background Machine learning (ML) methodology development for the classification of immune states in adaptive immune receptor repertoires (AIRRs) has seen a recent surge of interest. However, so far, there does not exist a systematic evaluation of scenarios where classical ML methods (such as penalized logistic regression) already perform adequately for AIRR classification. This hinders investigative reorientation to those scenarios where method development of more sophisticated ML approaches may be required. Results To identify those scenarios where a baseline ML method is able to perform well for AIRR classification, we generated a collection of synthetic AIRR benchmark data sets encompassing a wide range of data set architecture-associated and immune state–associated sequence patterns (signal) complexity. We trained ≈1,700 ML models with varying assumptions regarding immune signal on ≈1,000 data sets with a total of ≈250,000 AIRRs containing ≈46 billion TCRβ CDR3 amino acid sequences, thereby surpassing the sample sizes of current state-of-the-art AIRR-ML setups by two orders of magnitude. We found that L1-penalized logistic regression achieved high prediction accuracy even when the immune signal occurs only in 1 out of 50,000 AIR sequences. Conclusions We provide a reference benchmark to guide new AIRR-ML classification methodology by (i) identifying those scenarios characterized by immune signal and data set complexity, where baseline methods already achieve high prediction accuracy, and (ii) facilitating realistic expectations of the performance of AIRR-ML models given training data set properties and assumptions. Our study serves as a template for defining specialized AIRR benchmark data sets for comprehensive benchmarking of AIRR-ML methods.

• Akbar, Rahmad; Robert, Philippe Paul Auguste; Weber, Cédric R.; Widrich, Michael; Frank, Robert & Pavlovic, Milena [Vis alle 19 forfattere av denne artikkelen] (2022). In silico proof of principle of machine learning-based antibody design at unconstrained scale. mAbs. ISSN 1942-0862. 14:e2031482(1), s. 1–18. doi: 10.1080/19420862.2022.2031482. Fulltekst i vitenarkiv Vis sammendrag

  Generative machine learning (ML) has been postulated to become a major driver in the computational design of antigen-specific monoclonal antibodies (mAb). However, efforts to confirm this hypothesis have been hindered by the infeasibility of testing arbitrarily large numbers of antibody sequences for their most critical design parameters: paratope, epitope, affinity, and developability. To address this challenge, we leveraged a lattice-based antibody-antigen binding simulation framework, which incorporates a wide range of physiological antibody-binding parameters. The simulation framework enables the computation of synthetic antibody-antigen 3D-structures, and it functions as an oracle for unrestricted prospective evaluation and benchmarking of antibody design parameters of ML-generated antibody sequences. We found that a deep generative model, trained exclusively on antibody sequence (one dimensional: 1D) data can be used to design conformational (three dimensional: 3D) epitope-specific antibodies, matching, or exceeding the training dataset in affinity and developability parameter value variety. Furthermore, we established a lower threshold of sequence diversity necessary for high-accuracy generative antibody ML and demonstrated that this lower threshold also holds on experimental real-world data. Finally, we show that transfer learning enables the generation of high-affinity antibody sequences from low-N training data. Our work establishes a priori feasibility and the theoretical foundation of high-throughput ML-based mAb design.

• Yao, Ying; Zia, Asima; Neumann, Ralf Stefan; Pavlović, Milena; Balaban, Gabriel & Lundin, Knut Erik Aslaksen [Vis alle 8 forfattere av denne artikkelen] (2021). T cell receptor repertoire as a potential diagnostic marker for celiac disease. Clinical Immunology. ISSN 1521-6616. 222:108621, s. 1–7. doi: 10.1016/j.clim.2020.108621. Fulltekst i vitenarkiv Vis sammendrag

  An individual's T cell repertoire is skewed towards some specificities as a result of past antigen exposure and subsequent clonal expansion. Identifying T cell receptor signatures associated with a disease is challenging due to the overall complexity of antigens and polymorphic HLA allotypes. In celiac disease, the antigen epitopes are well characterised and the specific HLA-DQ2-restricted T-cell repertoire associated with the disease has been explored in depth. By investigating T cell receptor repertoires of unsorted lamina propria T cells from 15 individuals, we provide the first proof-of-concept study showing that it could be possible to infer disease state by matching against a priori known disease-associated T cell receptor sequences.

• Slabodkin, Andrei; Chernigovskaia, Maria; Mikocziova, Ivana; Akbar, Rahmad; Scheffer, Lonneke & Pavlović, Milena [Vis alle 16 forfattere av denne artikkelen] (2021). Individualized VDJ recombination predisposes the available Ig sequence space. Genome Research. ISSN 1088-9051. 31(12), s. 2209–2225. doi: 10.1101/gr.275373.121. Fulltekst i vitenarkiv Vis sammendrag

  The process of recombination between variable (V), diversity (D), and joining (J) immunoglobulin (Ig) gene segments determines an individual's naive Ig repertoire and, consequently, (auto)antigen recognition. VDJ recombination follows probabilistic rules that can be modeled statistically. So far, it remains unknown whether VDJ recombination rules differ between individuals. If these rules differed, identical (auto)antigen-specific Ig sequences would be generated with individual-specific probabilities, signifying that the available Ig sequence space is individual specific. We devised a sensitivity-tested distance measure that enables inter-individual comparison of VDJ recombination models. We discovered, accounting for several sources of noise as well as allelic variation in Ig sequencing data, that not only unrelated individuals but also human monozygotic twins and even inbred mice possess statistically distinguishable immunoglobulin recombination models. This suggests that, in addition to genetic, there is also nongenetic modulation of VDJ recombination. We demonstrate that population-wide individualized VDJ recombination can result in orders of magnitude of difference in the probability to generate (auto)antigen-specific Ig sequences. Our findings have implications for immune receptor–based individualized medicine approaches relevant to vaccination, infection, and autoimmunity.

• Arnaout, Ramy A.; Prak, Eline T. Luning; Schwab, Nicholas; Rubelt, Florian; Greiff, Victor & Pavlović, Milena [Vis alle 32 forfattere av denne artikkelen] (2021). The Future of Blood Testing Is the Immunome. Frontiers in Immunology. 12:626793, s. 1–6. doi: 10.3389/fimmu.2021.626793. Fulltekst i vitenarkiv Vis sammendrag

  It is increasingly clear that an extraordinarily diverse range of clinically important conditions—including infections, vaccinations, autoimmune diseases, transplants, transfusion reactions, aging, and cancers—leave telltale signatures in the millions of V(D)J-rearranged antibody and T cell receptor [TR per the Human Genome Organization (HUGO) nomenclature but more commonly known as TCR] genes collectively expressed by a person’s B cells (antibodies) and T cells. We refer to these as the immunome . Because of its diversity and complexity, the immunome provides singular opportunities for advancing personalized medicine by serving as the substrate for a highly multiplexed, near-universal blood test. Here we discuss some of these opportunities, the current state of immunome-based diagnostics, and highlight some of the challenges involved. We conclude with a call to clinicians, researchers, and others to join efforts with the Adaptive Immune Receptor Repertoire Community (AIRR-C) to realize the diagnostic potential of the immunome.

• Pavlović, Milena; Scheffer, Lonneke; Motwani, Keshav; Kanduri, Chakravarthi; Kompova, Radmila & Vazov, Nikolay Aleksandrov [Vis alle 42 forfattere av denne artikkelen] (2021). The immuneML ecosystem for machine learning analysis of adaptive immune receptor repertoires. Nature Machine Intelligence. 3(11), s. 936–944. doi: 10.1038/s42256-021-00413-z. Fulltekst i vitenarkiv
• Akbar, Rahmad; Robert, Philippe Paul Auguste; Pavlović, Milena; Jeliazkov, Jeliazko R.; Snapkov, Igor & Slabodkin, Andrei [Vis alle 15 forfattere av denne artikkelen] (2021). A compact vocabulary of paratope-epitope interactions enables predictability of antibody-antigen binding. Cell reports. 34:108856(11), s. 1–21. doi: 10.1016/j.celrep.2021.108856. Fulltekst i vitenarkiv Vis sammendrag

  Antibody-antigen binding relies on the specific interaction of amino acids at the paratope-epitope interface. The predictability of antibody-antigen binding is a prerequisite for de novo antibody and (neo-)epitope design. A fundamental premise for the predictability of antibody-antigen binding is the existence of paratope- epitope interaction motifs that are universally shared among antibody-antigen structures. In a dataset of non-redundant antibody-antigen structures, we identify structural interaction motifs, which together compose a commonly shared structure-based vocabulary of paratope-epitope interactions. We show that this vocabulary enables the machine learnability of antibody-antigen binding on the paratope-epitope level using generative machine learning. The vocabulary (1) is compact, less than 104 motifs; (2) distinct from non-immune protein-protein interactions; and (3) mediates specific oligo- and polyreactive interactions between paratope-epitope pairs. Our work leverages combined structure- and sequence-based learning to demonstrate that machine-learning-driven predictive paratope and epitope engineering is feasible.

• Weber, Cédric R.; Akbar, Rahmad; Yermanos, Alexander; Pavlovic, Milena; Snapkov, Igor & Sandve, Geir Kjetil [Vis alle 8 forfattere av denne artikkelen] (2020). immuneSIM: tunable multi-feature simulation of B- and T-cell receptor repertoires for immunoinformatics benchmarking. Bioinformatics. ISSN 1367-4803. 36(11), s. 3594–3596. doi: 10.1093/bioinformatics/btaa158. Fulltekst i vitenarkiv Vis sammendrag

  Abstract Summary B- and T-cell receptor repertoires of the adaptive immune system have become a key target for diagnostics and therapeutics research. Consequently, there is a rapidly growing number of bioinformatics tools for immune repertoire analysis. Benchmarking of such tools is crucial for ensuring reproducible and generalizable computational analyses. Currently, however, it remains challenging to create standardized ground truth immune receptor repertoires for immunoinformatics tool benchmarking. Therefore, we developed immuneSIM, an R package that allows the simulation of native-like and aberrant synthetic full-length variable region immune receptor sequences by tuning the following immune receptor features: (i) species and chain type (BCR, TCR, single and paired), (ii) germline gene usage, (iii) occurrence of insertions and deletions, (iv) clonal abundance, (v) somatic hypermutation and (vi) sequence motifs. Each simulated sequence is annotated by the complete set of simulation events that contributed to its in silico generation. immuneSIM permits the benchmarking of key computational tools for immune receptor analysis, such as germline gene annotation, diversity and overlap estimation, sequence similarity, network architecture, clustering analysis and machine learning methods for motif detection. Availability and implementation The package is available via https://github.com/GreiffLab/immuneSIM and on CRAN at https://cran.r-project.org/web/packages/immuneSIM. The documentation is hosted at https://immuneSIM.readthedocs.io. Contact sai.reddy@ethz.ch or victor.greiff@medisin.uio.no Supplementary information Supplementary data are available at Bioinformatics online.

• Brown, Alex; Snapkov, Igor; Akbar, Rahmad; Pavlovic, Milena; Miho, Enkelejda & Sandve, Geir Kjetil [Vis alle 7 forfattere av denne artikkelen] (2019). Augmenting adaptive immunity: progress and challenges in the quantitative engineering and analysis of adaptive immune receptor repertoires. Molecular Systems Design & Engineering. 4(4), s. 701–736. doi: 10.1039/c9me00071b. Fulltekst i vitenarkiv

Se alle arbeider i NVA

• Rand, Knut Dagestad; Grytten, Ivar; Pavlović, Milena; Kanduri, Chakravarthi & Sandve, Geir Kjetil Ferkingstad (2024). BioNumPy: array programming for biology. Nature Methods. ISSN 1548-7091. doi: 10.1038/s41592-024-02483-4. Fulltekst i vitenarkiv
• Torgersen, Eivind; Kjetil, Sandve Geir; Greiff, Victor; Pavlovic, Milena & Scheffer, Lonneke (2022). Fra bare én blodprøve kan kunstig intelligens gi diagnose for mange ulike sykdommer. Titan.uio.no. Fulltekst i vitenarkiv
• Widrich, Michael; Schäfl, Bernhard; Pavlović, Milena; Ramsauer, Hubert; Gruber, Lukas & Holzleitner, Markus [Vis alle 11 forfattere av denne artikkelen] (2020). Modern Hopfield Networks and Attention for Immune Repertoire Classification. Fulltekst i vitenarkiv Vis sammendrag

  A central mechanism in machine learning is to identify, store, and recognize patterns. How to learn, access, and retrieve such patterns is crucial in Hopfield networks and the more recent transformer architectures. We show that the attention mechanism of transformer architectures is actually the update rule of modern Hopfield networks that can store exponentially many patterns. We exploit this high storage capacity of modern Hopfield networks to solve a challenging multiple instance learning (MIL) problem in computational biology: immune repertoire classification. In immune repertoire classification, a vast number of immune receptors are used to predict the immune status of an individual. This constitutes a MIL problem with an unprecedentedly massive number of instances, two orders of magnitude larger than currently considered problems, and with an extremely low witness rate. Accurate and interpretable machine learning methods solving this problem could pave the way towards new vaccines and therapies, which is currently a very relevant research topic intensified by the COVID-19 crisis. In this work, we present our novel method DeepRC that integrates transformer-like attention, or equivalently modern Hopfield networks, into deep learning architectures for massive MIL such as immune repertoire classification. We demonstrate that DeepRC outperforms all other methods with respect to predictive performance on large-scale experiments including simulated and real-world virus infection data and enables the extraction of sequence motifs that are connected to a given disease class. Source code and datasets: https://github.com/ml-jku/DeepRC.

Se alle arbeider i NVA