Biology subjects

Dilip, R.

Publications and source records attributed to Dilip, R..

4 recordsLinked to original sources

bioRxiv2026

Automated assembly of protein complexes from cryo-EM maps with structure-informed Monte Carlo Tree Search

Structural cell biology aims to visualize functional molecules as they carry out their biological roles in their native cellular context. However, macromolecular complexes in situ have thus far been resolved predominantly at intermediate resolutions, complicating protein identification and structural modeling due to the vast combinatorial space of possible components within a proteome. Here, we developed Cryosearch, a GPU-accelerated framework for automated assembly of macromolecular complexes from proteome-scale monomer libraries. Cryosearch implements Monte Carlo tree search with correlation-based rewards to identify combinations of protein domains that collectively best explain density maps. This approach enables autonomous de novo assembly of molecular complexes from intermediate-resolution maps.

bioinformatics↗

bioRxiv2025

Flow Autoencoders are Effective Protein Tokenizers

AO_SCPLOWBSTRACTC_SCPLOWProtein structure tokenizers enable the creation of multimodal models of protein structure, sequence, and function. Current approaches to protein structure tok-enization rely on bespoke components that are invariant to spatial symmetries, but that are challenging to optimize and scale. We present Kanzi, a flow-based tokenizer for tokenization and generation of protein structures. Kanzi consists of a diffusion autoencoder trained with a flow matching loss. We show that this approach simplifies several aspects of protein structure tokenizers: frame-based representations can be replaced with global coordinates, complex losses are replaced with a single flow matching loss, and SE(3)-invariant attention operations can be replaced with standard attention. We find that these changes stabilize the training of parameter-efficient models that outperform existing to- kenizers on reconstruction metrics at a fraction of the model size and training cost. An autoregressive model trained with Kanzi outperforms similar generative models that operate over tokens, although it does not yet match the performance of state-of-the-art continuous diffusion models. Code is available here: https://github.com/rdilip/kanzi/.

bioengineering↗

bioRxiv2024

Generalized cell phenotyping for spatial proteomics with language-informed vision models

We present DeepCell Types, a novel approach to cell phenotyping for spatial proteomics that addresses the challenge of generalization across diverse datasets with varying marker panels collected across different platforms. Our approach utilizes a transformer with channel-wise attention to create a language-informed vision model; this models semantic understanding of the underlying marker panel enables it to learn from and adapt to heterogeneous datasets. Leveraging a curated, diverse dataset named Expanded TissueNet with cell type labels spanning the literature and the NIH Human BioMolecular Atlas Program (HuBMAP) consortium, our model demonstrates robust performance across various cell types, tissues, and imaging modalities. Comprehensive benchmarking shows that our method outperforms existing approaches on cell-type prediction and, from the same model, predicts marker positivity competitively with a dedicated specialist; it further matches manual expert gating and adapts to new data with modest fine-tuning, well past what baselines reach when trained from scratch. This work equips the spatial proteomics community with a single, continuously improvable phenotyping model that generalizes to new marker panels and can be fine-tuned efficiently when needed. We release both DeepCell Types and Expanded TissueNet as open-source resources.

bioinformatics↗

bioRxiv2023

A Foundation Model for Cell Segmentation

Cells are a fundamental unit of biological organization, and identifying them in imaging data - cell segmentation - is a critical task for various cellular imaging experiments. While deep learning methods have led to substantial progress on this problem, most models are specialist models that work well for specific domains but cannot be applied across domains or scale well with large amounts of data. In this work, we present CellSAM, a universal model for cell segmentation that generalizes across diverse cellular imaging data. CellSAM builds on top of the Segment Anything Model (SAM) by developing a prompt engineering approach for mask generation. We train an object detector, CellFinder, to automatically detect cells and prompt SAM to generate segmentations. We show that this approach allows a single model to achieve human-level performance for segmenting images of mammalian cells, yeast, and bacteria collected across various imaging modalities. We show that CellSAM has strong zero-shot performance and can be improved with a few examples via few-shot learning. Additionally, we demonstrate how CellSAM can be applied across diverse bioimage analysis workflows. A deployed version of CellSAM is available at https://cellsam.deepcell.org/.

bioinformatics↗