The intricate dance of DNA to RNA to protein is the foundation of life, a process first unveiled with Watson and Crick's 1953 double helix model. Yet, after translation, proteins must fold into precise conformations and undergo a symphony of post-translational modifications (PTMs), chemical tweaks like phosphorylation, glycosylation, and ubiquitination, that dictate their function, stability, and interactions. These PTMs are a protein's code, controlling cellular processes from signaling to degradation. In cystic fibrosis (CF), a devastating genetic disorder affecting 70,000 people worldwide, dysfunctional PTMs on the CFTR protein disrupt chloride transport, leading to thick mucus buildup in lungs and pancreas.
At the forefront of this research is the Yates Laboratory at The Scripps Research Institute in La Jolla, California, led by renowned mass spectrometry pioneer John Yates. Funded by the NIH (Grant ID: 1R01HL131697-01A1), the lab, featuring researchers Robin Park, Sandra Pankow, Titus Jung, and others, is unraveling how PTMs influence CFTR's misfolding in the ΔF508 mutation, the most common CF cause. By partnering with Omnibond®'s advanced cloud HPC technology, they've transformed a computationally intensive "blind PTM search" from weeks of gridlock into hours of discovery, identifying novel modifications that could unlock new therapies.
The Challenge: Decoding PTMs in Low-Abundance CFTR
CF arises from mutations in the CFTR gene, with 90% of cases linked to ΔF508, a deletion of phenylalanine at position 508. This alters CFTR's 3D structure, preventing proper folding and membrane trafficking, crippling its anion channel function. Initial Yates Lab studies revealed that post-translational modifications on misfolded ΔF508 CFTR differ from wild-type (wt-CFTR), hinting at regulatory roles in disease pathology.
Identifying CFTR PTMs is crucial for understanding channel defects and developing therapeutics, but it's daunting:
- Low Protein Abundance: CFTR levels in cells are extremely low, making detection challenging amid cellular noise.
- Modification Diversity: Hundreds of PTM types exist, with combinations per protein numbering in the thousands, many below detection thresholds.
- Computational Bottleneck: Traditional searches against known PTMs miss novel ones; unbiased "blind PTM" scans against the entire human genome are exhaustive, taking up to two weeks on a 600-node cluster for a single experiment.
The Yates Lab needed an unbiased approach to reveal all PTMs (including SNPs and deletions) without bias toward predicted molecular weights, signaling the presence of modifications through peptide sequence deviations. However, blind PTM searches are computationally demanding, such queries can overwhelm even large clusters.
Omnibond®'s Solution: Cloud HPC Enables Parallelized Blind PTM Discovery
Omnibond® collaborated with the Yates Lab to deploy our cutting-edge cloud HPC technology, integrating seamlessly with their IP2 (Integrated Proteomic Pipeline) workflow. This end-to-end system, from sample login to elastic compute jobs, transforms mass spectrometry data into actionable insights, now supercharged for blind PTM searches.
- Elastic Compute Scaling: Our provisioning engine launches parallel jobs across cloud instances, handling login, CCD (charge state determination), and SLURM queuing. For blind PTM, it distributes genome-wide searches across thousands of vCPUs, reducing run times from weeks to hours on a single experiment.
- Cloud Parallel Storage: Integrates with object storage (e.g., GCS or S3) for massive datasets, enabling VPC-secured, high-throughput IO without bottlenecks. This supports the 15,000-20,000 spectra per GB generated by LTQ-Orbitrap Elite instruments.
- Hybrid Workflow Federation: CCQ meta-scheduler overlays SLURM, preventing overload from massive queries while optimizing for preemptible instances. It federates with on-prem systems, allowing seamless bursts without retooling.
Omnibond®'s expertise customized IP2 for high-throughput proteomics, used by labs, universities, pharma companies, and hospitals worldwide. As John Yates notes, "Our cloud integration unlocks unprecedented scale for unbiased searches, revealing PTMs that traditional methods miss."
Data & Results: Revealing Novel PTMs in ΔF508 CFTR
Over 90 independent samples were analyzed on high-accuracy mass spectrometers like the LTQ-Orbitrap Elite (Thermo Fisher). Each experiment generated >1 GB of data, about 15,000-20,000 spectra identified by traditional ProLuCID searches.
- Unbiased Identification: Traditional ProLuCID identified PTMs around 80,000 CFTR spectra, revealing important alterations in cystic fibrosis. Blind PTM uncovered an additional 30,000 CFTR spectra with various PTMs.
- Critical Insights: The blind search revealed PTM changes that would have been missed by traditional searches, providing crucial information on how PTMs influence ΔF508 CFTR defects.
- Therapeutic Potential: The Yates Lab is following up on several PTMs to see if influencing some could aid CF therapy; a paper describing the results will be published soon.
Quantitative and qualitative analysis showed differentially altered PTMs in CF, highlighting regulatory roles in misfolding and trafficking. This unbiased approach, peptide sequences identified along with deviations from predicted weights, signals novel modifications invisible to targeted methods.
Outcomes: Accelerating CF Therapies Through Cloud-Enabled Discovery
Omnibond®'s cloud HPC slashed blind PTM run times dramatically, enabling higher parallelization and significant reductions in compute costs. This empowers the Yates Lab to process massive datasets affordably, uncovering PTMs that could target CFTR rescue in ΔF508 mutants, potentially restoring channel function.
The IP2 workflow's scalability extends beyond CF: It's now used by diverse research labs, universities, pharmaceutical companies, and hospitals for unbiased proteomics. As Titus Jung explains, "CloudyCluster's elastic bursting turns weeks of waiting into hours of insight, democratizing PTM discovery."
How this relates to projectEureka™
This work is an Omnibond® company capability in cloud HPC for biomedical research. It shows the kind of long-running, data-heavy science projectEureka is designed to host in project-centric workspaces, not that every PTM pipeline is a built-in product feature. Researchers still bring their codes and workflows; the platform provides the shared environment, apps, and storage next to the work.
Sources: Based on Yates Lab slides on blind PTM search for CFTR, highlighting Omnibond's cloud HPC integration and IP2 workflow.