Work
Research
Publications, talks and courses. The work is on protein language models, gene regulation and multi-omic data integration.
Topics
- The language of life: protein language models, gene regulation and multi-omic data integration.
- Variant effect prediction and functional annotation of protein sequence.
- Measuring diversity with document embeddings: product diversity and open-ended innovation.
- Biology at the bench: group testing for pathogen screening and the stress response of industrial yeast.
Publications
- Unified imputation of missing data modalities and features in multi-omic data via shared representation learning
bioRxiv · 2026-02 · Preprint
Preprint: one shared representation for several omic layers, so a missing modality or feature can be imputed from the others instead of being dropped.
- BPS2026 - A brighter future: Optimizing GFP fluorescence using deep learning
Biophysical Journal · 2026-02 · Paper
Meeting abstract: a deep learning approach to optimizing GFP fluorescence, the work Ananthan did with the Anderson lab at the Biophysical Society meeting.
- Protein Language Models Capture Structural and Functional Epistasis in a Zero-Shot Setting
bioRxiv · 2025-09 · Preprint
Preprint: epistasis as a lens on what protein language models learn. Raw PLM scores align with residue-residue contacts, and a nonlinear transformation shifts the signal toward functional couplings between distant sites.
- Machine Learning for Protein Science and Engineering
Cold Spring Harbor Perspectives in Biology · 2025-05 · Editorial
A collection on machine learning for protein science, edited with Peter K. Koo, Christian Dallago and Kevin K. Yang: representations, co-evolutionary statistical models, homology search, functional annotation and variant effect prediction.
- DR-BERT: A protein language model to annotate disordered regions
Structure · 2024-08 · Paper
A compact protein language model that predicts intrinsically disordered regions without relying on evolutionary or biophysical data, outperforming existing methods on the CAID evaluation datasets.
- Semi-quantitative group testing for efficient and accurate qPCR screening of pathogens with a wide range of loads
BMC Bioinformatics · 2024-05 · Paper
A group testing scheme for qPCR pathogen screening that handles viral loads spanning several orders of magnitude, with theoretical guarantees on the number of tests required.
- A transcriptomic atlas of acute stress response to low pH in multiple Issatchenkia orientalis strains
Microbiology Spectrum · 2024-01 · Paper
A transcriptomic atlas of the low-pH stress response across multiple strains of the acid-tolerant yeast Issatchenkia orientalis, a chassis for industrial organic acid production.
- FUN-PROSE: A deep learning approach to predict condition-specific gene expression in fungi
PLOS Computational Biology · 2023-11 · Paper
A deep learning model that predicts differential expression of individual genes across conditions from promoter sequences and transcription factor expression, with an interpretable layer that recovers promoter motifs and TF-gene interactions.
- Transformer Neural Networks for Protein Family and Interaction Prediction Tasks
Journal of Computational Biology · 2023-01 · Paper
Journal version of the ACM-BCB 2020 paper: a small transformer pretrained on protein sequences, fine-tuned for protein family classification and protein-protein interaction prediction at a fraction of the compute of larger models.
- Dropping diversity of products of large US firms: Models and measures
PLOS ONE · 2022-03 · Paper
Measures and models of dropping product diversity in large US firms, built on document embeddings of firm filings.
- Transforming the Language of Life: Transformer Neural Networks for Protein Prediction Tasks
ACM-BCB 2020 · 2020-09 · Paper
Conference paper: a small transformer pretrained on protein sequence representations, fine-tuned to state-of-the-art performance on protein family classification and interaction prediction. The first paper in the language-of-life line.
- Open-Ended Technological Innovation
Artificial Life · 2019-04 · Paper
A framework for detecting open-ended growth in the space of technological innovations, applied to patent records.
Talks
Transcribing the Language of Life: An Interpretable Deep Neural Network to Predict Condition Specific Gene Regulation
ISMB/ECCB 2021 · 2021-07 · Poster
Poster. The interpretable neural network behind FUN-PROSE, which predicts condition-specific gene regulation and recovers the promoter motifs responsible.
Computing the Language of Life: NLP Approaches to Feature Extraction for Protein Family Classification
ISMB/ECCB 2019 · 2019-07 · Poster
Poster. Feature extraction for protein family classification, treating protein sequences as language.
Detecting the On-going Emergence of Technological Innovations
ALIFE 2018, Workshop on Open-ended Evolution · 2018-07 · Talk
Oral presentation. Detection of ongoing technological innovation, later published as Open-Ended Technological Innovation in Artificial Life.
Controllability of Functional Classes in the Genetic Regulatory Network of E. coli
SIAM Annual Meeting 2018 · 2018-07 · Poster
Poster. Target controllability of functional classes of genes in E. coli, from the network biology work done at Delhi University and the Santa Fe Institute.