Putting computational models of immunity to the test - an invited challenge to predict B. pertussis vaccination outcomes.

Systems vaccinology studies have been used to build computational models that predict individual vaccine responses and identify the factors contributing to differences in outcome. Comparing such models is challenging due to variability in study designs. To address this, we established a community resource to compare models predicting B. pertussis booster responses and generate experimental data for the explicit purpose of model evaluation. We here describe our second computational prediction challenge using this resource, where we benchmarked 49 algorithms from 53 scientists. We found that the most successful models stood out in their handling of nonlinearities, reducing large feature sets to representative subsets, and advanced data preprocessing. In contrast, we found that models adopted from literature that were developed to predict vaccine antibody responses in other settings performed poorly, reinforcing the need for purpose-built models. Overall, this demonstrates the value of purpose-generated datasets for rigorous and open model evaluations to identify features that improve the reliability and applicability of computational models in vaccine response prediction.

Enhancing personalized gene expression prediction from DNA sequences using genomic foundation models.

Artificial intelligence (AI)/deep learning (DL) models that predict molecular phenotypes like gene expression directly from DNA sequences have recently emerged. While these models have proven effective at capturing the variation across genes, their ability to explain inter-individual differences has been limited. We hypothesize that the performance gap can be narrowed through the use of pre-trained embeddings from the Nucleotide Transformer, a large foundation model trained on 3,000+ genomes. We train a transformer model using the pre-trained embeddings and compare its predictive performance to Enformer, the current state-of-the-art model, using genotype and expression data from 290 individuals. Our model significantly outperforms Enformer in terms of correlation across individuals, and narrows the performance gap with an elastic net regression approach that uses just the genetic variants as predictors. Although simple regression models have their advantages in personalized prediction tasks, DL approaches based on foundation models pre-trained on diverse genomes have unique strengths in flexibility and interpretability. With further methodological and computational improvements with more training data, these models may eventually predict molecular phenotypes from DNA sequences with an accuracy surpassing that of regression-based approaches. Our work demonstrates the potential for large pre-trained AI/DL models to advance functional genomics.

Molecular predictors of immunophenotypic measurable residual disease clearance in acute myeloid leukemia.

Measurable residual disease (MRD) is a powerful prognostic factor in acute myeloid leukemia (AML). However, pre-treatment molecular predictors of immunophenotypic MRD clearance remain unclear. We analyzed a dataset of 211 patients with pre-treatment next-generation sequencing who received induction chemotherapy and had MRD assessed by serial immunophenotypic monitoring after induction, subsequent therapy, and allogeneic stem cell transplant (allo-SCT). Induction chemotherapy led to MRD- remission, MRD+ remission, and persistent disease in 35%, 27%, and 38% of patients, respectively. With subsequent therapy, 34% of patients with MRD+ and 26% of patients with persistent disease converted to MRD-. Mutations in CEBPA, NRAS, KRAS, and NPM1 predicted high rates of MRD- remission, while mutations in TP53, SF3B1, ASXL1, and RUNX1 and karyotypic abnormalities including inv (3), monosomy 5 or 7 predicted low rates of MRD- remission. Patients with fewer individual clones were more likely to achieve MRD- remission. Among 132 patients who underwent allo-SCT, outcomes were favorable whether patients achieved early MRD- after induction or later MRD- after subsequent therapy prior to allo-SCT. As MRD conversion with chemotherapy prior to allo-SCT is rarely achieved in patients with specific baseline mutational patterns and high clone numbers, upfront inclusion of these patients into clinical trials should be considered.