Suresh Marru

Current AI protein structure prediction models involve multi-stage processing that combines deep neural networks with bioinformatics tools such as multiple sequence alignment (MSA). Researchers increasingly rely on intermediate or penultimate-layer activations from these models for downstream tasks including contact prediction, binding-site identification, and model interpretability. We describe the VizFold plugin, a modular framework that can be extended toward end-to-end composable pipelines. We demonstrate feasibility through standardized hook-based tracing for ESMFold and Boltz-2, archive validation, and reproducible deployment on an HPC cluster using managed caches, modules, quotas, and Slurm workflows. The framework extracts attention maps from user-selected layers and exports intermediate representations in a backend-specific run bundle (Boltz) or a canonical archive tree (ESMFold), with shared trace text conventions and explicit provenance metadata suitable for cross-model comparison. We provide step-by-step documentation for instrumentation and deployment so that other groups can reproduce or extend the pipeline on their own clusters. The framework and instrumentation code are open-source and available at https://github.com/AI2Science/vizfold-foundation.

Scientists often utilize personal laptops and workstations for initial research stages and turn to high-performance computing (HPC) supercomputers for compute-intensive tasks. However, seamless transitions between these environments are vital for enhancing productivity and accelerating research progress. Our paper presents the Cybershuttle Notebook Gateway, an open-source framework crafted to streamline this transition, optimize resource utilization, and reduce time-to-science for researchers. Leveraging JupyterLab, the framework extends kernel mechanics for seamless provisioning and connection to remote HPC cluster kernels. We delve into its architecture, which separates user authentication, kernel provisioning, and remote file system access. Additionally, we highlight practical capabilities like analyzing network load and CPU usage across various computational workloads.

Software-as-a-service science gateways provide user interfaces and middleware for accessing scientific software deployed on remote high-performance computing resources and clusters. Selecting the resource to use for a particular job submission may be left to the user, who may need more information to make good choices when selecting from multiple options. To address this problem, we have designed and developed an extensible, scalable metascheduling system that can provide automated scheduling capabilities based on resource availability and other characteristics. We develop a system model based on queuing theory to guide our implementation and provide a basis for analysis. In particular, we derive an efficiency metric from these considerations. We implement the metascheduling system within the open-source Apache Airavata framework for science gateways as a supplemental service for guiding the job submission capabilities. We measure efficiency in representative scenarios, observing efficiencies of greater than 70% even in scenarios with high input rates and low job acceptance rates.

Metadata catalogs are essential for enabling researchers to find and access relevant datasets. However, existing metadata catalog solutions have limitations, such as being domain-specific or using document-oriented databases that limit their scalability and flexibility. To address these issues, we introduce the Airavata Data Catalog - a multi-tenanted schema-free metadata catalog service that supports multiple domain-specific metadata schemas and access control mechanisms. This paper describes our approach to modeling the relational attributes of data products and their access control while supporting a schema-free, document-oriented approach to storing and searching metadata. Our approach offers significant improvements over existing solutions and demonstrates the feasibility of a scalable, flexible metadata catalog for scientific datasets.

This paper introduces Cybershuttle, a new type of user-facing cyberinfrastructure that provides seamless access to a range of resources for researchers, enhancing their productivity. The Cybershuttle Research Environment is built on open source Apache Airavata software and uses a hybrid approach that integrates locally deployed agent programs with centrally hosted middleware. This enables end-to-end integration of computational science and engineering research across a range of resources, including users’ local resources, centralized university computing and data resources, computational clouds, and NSF-funded national-scale computing centers. To ensure a user-centered approach, we have designed the scientific user environments with the best user-centered design practices.

Jupyter Notebooks are an enormously popular tool for creating and narrating computational research projects. They also have enormous potential for creating reproducible scientific research artifacts. Capturing the complete state of a notebook has additional benefits; for instance, the notebook execution may be split between local and remote resources, where the latter may have more powerful processing capabilities or store large or access-limited data. There are several challenges for making notebooks fully reproducible when examined in detail. The notebook code must be replicated entirely, and the underlying Python runtime environments must be identical. More subtle problems arise in replicating referenced data, external library dependencies, and runtime variable states. This paper presents solutions to these problems using Juptyer's standard extension mechanisms to create an archivable system state for a running notebook. We show that the overhead for these additional mechanisms, which involve interacting with the underlying Linux kernel, does not introduce substantial execution time overheads, demonstrating the approach's feasibility.

Top-down mass spectrometry-based proteomics has become the method of choice for identifying and quantifying intact proteoforms in biological samples. We present a web-based gateway for TopPIC suite, a widely used software suite consisting of four software tools for top-down mass spectrometry data interpretation: TopFD, TopPIC, TopMG, and TopDiff. The gateway enables the community to use heterogeneous collection of computing resources that includes high performance computing clusters at Indiana University and virtual clusters on XSEDE’s Jetstream Cloud resource for top-down mass spectral data analysis using TopPIC suite. The gateway will be a useful resource for proteomics researchers and students who have limited access to high-performance computing resources or who are not familiar with interacting with server-side supercomputers.

The amount of DNA sequencing data has been exponentially growing during the past decade due to advances in sequencing technology. Processing and modeling large amounts of sequencing data can be computationally intractable for desktop computing platforms. High performance computing (HPC) resources offer advantages in terms of computing power, and can be a general solution to these problems. Using HPCs directly for computational needs requires skilled users who know their way around HPCs and acquiring such skills take time. Science gateways acts as the middle layer between users and HPCs, providing users with the resources to accomplish compute-intensive tasks without requiring specialized expertise. We developed a web-based computing platform for genome biologists by customizing the PHP Gateway for Airavata (PGA) framework that accesses publicly accessible HPC resources via Apache Airavata. This web computing platform takes advantage of the Extreme Science and Engineering Discovery Environment (XSEDE) which provides the resources for gateway development, including access to CPU, GPU, and storage resources. We used this platform to develop a gateway for the dREG algorithm, an online computing tool for finding functional regions in mammalian genomes using nascent RNA sequencing data. The dREG gateway provides its users a free, powerful and user-friendly GPU computing resource based on XSEDE, circumventing the need of specialized knowledge about installation, configuration, and execution on an HPC for biologists. The dREG gateway is available at: https://dREG.dnasequence.org/.