Geir Kjetil Sandve

Milena Pavlović, Ghadi S. Al Hajj, Chakravarthi Kanduri et al.

Machine learning is increasingly used to discover diagnostic and prognostic biomarkers from high-dimensional molecular data. However, a variety of factors related to experimental design may affect the ability to learn generalizable and clinically applicable diagnostics. Here, we argue that a causal perspective improves the identification of these challenges and formalizes their relation to the robustness and generalization of machine learning-based diagnostics. To make for a concrete discussion, we focus on a specific, recently established high-dimensional biomarker - adaptive immune receptor repertoires (AIRRs). Through simulations, we illustrate how major biological and experimental factors of the AIRR domain may influence the learned biomarkers. In conclusion, we argue that causal modeling improves machine learning-based biomarker robustness by identifying stable relations between variables and by guiding the adjustment of the relations and variables that vary between populations.

Mai Ha Vu, Rahmad Akbar, Philippe A. Robert et al.

Deep neural-network-based language models (LMs) are increasingly applied to large-scale protein sequence data to predict protein function. However, being largely black-box models and thus challenging to interpret, current protein LM approaches do not contribute to a fundamental understanding of sequence-function mappings, hindering rule-based biotherapeutic drug development. We argue that guidance drawn from linguistics, a field specialized in analytical rule extraction from natural language data, can aid with building more interpretable protein LMs that are more likely to learn relevant domain-specific rules. Differences between protein sequence data and linguistic sequence data require the integration of more domain-specific knowledge in protein LMs compared to natural language LMs. Here, we provide a linguistics-based roadmap for protein LM pipeline choices with regard to training data, tokenization, token embedding, sequence embedding, and model interpretation. Incorporating linguistic ideas into protein LMs enables the development of next-generation interpretable machine-learning models with the potential of uncovering the biological mechanisms underlying sequence-function relationships.

Mai Ha Vu, Philippe A. Robert, Rahmad Akbar et al.

Apparent parallels between natural language and biological sequence have led to a recent surge in the application of deep language models (LMs) to the analysis of antibody and other biological sequences. However, a lack of a rigorous linguistic formalization of biological sequence languages, which would define basic components, such as lexicon (i.e., the discrete units of the language) and grammar (i.e., the rules that link sequence well-formedness, structure, and meaning) has led to largely domain-unspecific applications of LMs, which do not take into account the underlying structure of the biological sequences studied. A linguistic formalization, on the other hand, establishes linguistically-informed and thus domain-adapted components for LM applications. It would facilitate a better understanding of how differences and similarities between natural language and biological sequences influence the quality of LMs, which is crucial for the design of interpretable models with extractable sequence-functions relationship rules, such as the ones underlying the antibody specificity prediction problem. Deciphering the rules of antibody specificity is crucial to accelerating rational and in silico biotherapeutic drug design. Here, we formalize the properties of the antibody language and thereby establish not only a foundation for the application of linguistic tools in adaptive immune receptor analysis but also for the systematic immunolinguistic studies of immune receptor specificity in general.

Ghadi S. Al Hajj, Johan Pensar, Geir Kjetil Sandve

Data simulation is fundamental for machine learning and causal inference, as it allows exploration of scenarios and assessment of methods in settings with full control of ground truth. Directed acyclic graphs (DAGs) are well established for encoding the dependence structure over a collection of variables in both inference and simulation settings. However, while modern machine learning is applied to data of an increasingly complex nature, DAG-based simulation frameworks are still confined to settings with relatively simple variable types and functional forms. We here present DagSim, a Python-based framework for DAG-based data simulation without any constraints on variable types or functional relations. A succinct YAML format for defining the simulation model structure promotes transparency, while separate user-provided functions for generating each variable based on its parents ensure simulation code modularization. We illustrate the capabilities of DagSim through use cases where metadata variables control shapes in an image and patterns in bio-sequences.

Michael Widrich, Bernhard Schäfl, Hubert Ramsauer et al.

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. 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. Immune repertoire classification based on the vast number of immunosequences of an individual is 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. 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

Hubert Ramsauer, Bernhard Schäfl, Johannes Lehner et al.

We introduce a modern Hopfield network with continuous states and a corresponding update rule. The new Hopfield network can store exponentially (with the dimension of the associative space) many patterns, retrieves the pattern with one update, and has exponentially small retrieval errors. It has three types of energy minima (fixed points of the update): (1) global fixed point averaging over all patterns, (2) metastable states averaging over a subset of patterns, and (3) fixed points which store a single pattern. The new update rule is equivalent to the attention mechanism used in transformers. This equivalence enables a characterization of the heads of transformer models. These heads perform in the first layers preferably global averaging and in higher layers partial averaging via metastable states. The new modern Hopfield network can be integrated into deep learning architectures as layers to allow the storage of and access to raw input data, intermediate results, or learned prototypes. These Hopfield layers enable new ways of deep learning, beyond fully-connected, convolutional, or recurrent networks, and provide pooling, memory, association, and attention mechanisms. We demonstrate the broad applicability of the Hopfield layers across various domains. Hopfield layers improved state-of-the-art on three out of four considered multiple instance learning problems as well as on immune repertoire classification with several hundreds of thousands of instances. On the UCI benchmark collections of small classification tasks, where deep learning methods typically struggle, Hopfield layers yielded a new state-of-the-art when compared to different machine learning methods. Finally, Hopfield layers achieved state-of-the-art on two drug design datasets. The implementation is available at: https://github.com/ml-jku/hopfield-layers

Robert Frank, Michael Widrich, Rahmad Akbar et al.

Generative machine learning models offer a powerful framework for therapeutic design by efficiently exploring large spaces of biological sequences enriched for desirable properties. Unlike supervised learning methods, which require both positive and negative labeled data, generative models such as LSTMs can be trained solely on positively labeled sequences, for example, high-affinity antibodies. This is particularly advantageous in biological settings where negative data are scarce, unreliable, or biologically ill-defined. However, the lack of attribution methods for generative models has hindered the ability to extract interpretable biological insights from such models. To address this gap, we developed Generative Attribution Metric Analysis (GAMA), an attribution method for autoregressive generative models based on Integrated Gradients. We assessed GAMA using synthetic datasets with known ground truths to characterize its statistical behavior and validate its ability to recover biologically relevant features. We further demonstrated the utility of GAMA by applying it to experimental antibody-antigen binding data. GAMA enables model interpretability and the validation of generative sequence design strategies without the need for negative training data.