Solving the Puzzle Faster: How We’re Helping Scientists Decode Disordered Proteins with SpinForecast
September 21, 2026
Bind
Bind Research Blog
SpinForecast is an open-access tool that accelerates NMR backbone assignment for intrinsically disordered proteins and regions, helping researchers move from raw spectra to biological insight and decision-making faster. Built by Bind Research in collaboration with University College London, SpinForecast helps remove one of the biggest bottlenecks in characterising challenging proteins central to health and disease.
Disordered proteins are challenging to study
Proteins drive the vast majority of biological processes, in all living systems. Understanding the 3D structure of proteins provides critical insights into their function. These insights support broad applications, including drug discovery in biopharma, improved crop fitness in agriculture, and enzyme engineering for industrial biocatalysis.
Roughly 70% of the proteome consists of proteins with well-defined 3D structures. In general, clear representations of their structure can be obtained with tools such as X-ray crystallography, cryo-electron microscopy (cryo-EM), and nuclear magnetic resonance (NMR) spectroscopy.
The remaining 30% of the proteome is composed of intrinsically disordered proteins (IDPs) and proteins with intrinsically disordered regions (IDRs), that never settle into a single fixed structure; instead, they sample an ensemble of conformations. Just like photographing a Labrador mid-sprint, an IDP/IDR often returns a blurred, uninformative representation when studied with cryo-EM or X-ray crystallography. This makes IDPs/IDRs notoriously difficult to study, despite their well-appreciated roles in healthy cellular signalling, disease progression, and stress adaptation.
Backbone assignments are a bottleneck for IDP/IDR characterization
NMR is the only technique in the protein characterisation toolbox that can detect extreme motion, and therefore can accurately represent the true conformational ensemble of an IDP/IDR in solution. NMR works by detecting how atomic nuclei in the protein (e.g., hydrogen, carbon, nitrogen) respond in a magnetic field, producing a spectrum made up of signals that reflect the local chemical environment of each residue within the protein backbone. The raw spectrum, however, does not inform the viewer which signals correspond to which atoms.
Backbone assignment is the process of matching signals in an NMR spectrum to the correct atoms and residues in the protein sequence. Until those matches are made, the spectrum remains a pattern of signals rather than actionable biological insight. This step involves tracing the amino acid chain by analysing the chemical signatures of a signal and those of its neighbours. It is typically slow, labour intensive, and often the rate-limiting step in the workflow. Backbone assignment is unglamorous work and a key bottleneck standing between a purified protein in a tube and actionable insight that supports publications, grant applications and/or strategic decisions.
SpinForecast relieves the backbone assignment bottleneck
SpinForecast, developed by Bind Research in collaboration with University College London, is an open-access tool that performs backbone assignment without ever building a chain, by collapsing weeks of manual work into a task that can be conducted computationally in less than a minute.
To build SpinForecast, we used statistical patterns learned from previously studied IDPs/IDRs to predict the position on the spectrum of a peak corresponding to a specific amino acid in a specific sequence context. We mined the Biomolecular Magnetic Resonance Data Bank (BMRB), the world’s central archive of biological NMR data, and used artificial intelligence (AI) structure predictions from AlphaFold2 to isolate the predicted disordered stretches within it.
From that dataset, SpinForecast builds residue-specific fingerprints that account for sequence context and experimental conditions such as temperature and pH. When a new spectrum is analysed, the tool asks a simple question: given this signal’s position, which residue is it most likely to come from? Because every signal is evaluated independently of its location in the chain, a missing or ambiguous neighbour does not derail the whole backbone assignment (see Figure 1). This is particularly valuable for IDPs/IDRs, where overlapping peaks are a frequent occurrence.
In addition to accelerating the backbone assignment process, SpinForecast can be applied to validate manually-generated assignments for signals arising from minor populations, which often cannot be detected with chain-based methods.
Proof-of-concept of accelerated backbone assignment
We tested SpinForecast on independently determined assignments for three disordered proteins of increasing size and complexity. Across the purely disordered proteins, SpinForecast made 100% confident assignments for ~40-75% of signals, with an accuracy greater than 99%. For the remaining signals, it returned a ranked shortlist of likely candidates rather than a single guess; the correct residue was on that shortlist over 97% of the time.
We compared, side-by-side, the time savings of a workflow where we incorporated SpinForecast to generate the backbone assignments relative to a fully manual approach for two of these proteins. Depending on the protein length, incorporation of SpinForecast into the workflow resulted in time savings ranging from days to over a month (Figure 2).
Figure 2: SpinForecast accelerates backbone assignments of intrinsically disordered proteins and regions.
Workflow incorporation of SpinForecast
SpinForecast helps anchor known points across a sequence so the remaining gaps can be filled with traditional methods more efficiently, drastically reducing the timelines for backbone assignment (Figure 3).
For industry scientists and decision makers, these time savings can convert into accelerated timelines for key scientific and strategic milestones.
For academic NMR users, SpinForecast can lower the entry barrier for studying disordered proteins and speed up the path from raw spectra to insight, publication and/or grant submission.
Above all, we built SpinForecast to support a larger mission: expanding access to advanced NMR analysis in an area of high unmet need. Like all our tools, SpinForecast is open-access, user-friendly, and built to help researchers answer questions that were previously too challenging to address. You can use SpinForecast online or by installing the source code on your local machine.
Thousands of disordered proteins implicated in disease remain uncharacterised at atomic resolution. Backbone assignment by NMR has simply been too slow and too costly to unlock them. Our mission is to change that. SpinForecast is the first tool from Bind Research designed to make IDPs druggable for everyone.
Tell us how you use SpinForecast to advance your project by leaving a message here: https://bindresearch.org/contact/
Learn more about SpinForecast in this preprint: https://www.biorxiv.org/content/10.64898/2026.07.31.740808v1
Use the tool: tools.bindresearch.org/bindbox/SpinForecast
Explore the underlying NMR data: tools.bindresearch.org/bindbox/BMRB_Chemical_Shifts
Read the source code: github.com/bindresearch/BindBox