Werner Braun
University of Texas Medical Branch at Galveston, TX·Biochemistry
What makes one protein trigger an allergic reaction while a nearly identical protein is harmless? Allergies are among the most common human disorders in developed countries.
- Fields
Biochemistry
Microbiology & Immunology
Computer Science
- Topics
allergens
food allergy
peanut allergy
- Grants
- 2 known grants, through Jan 2031
What they work on
- Add 3D protein structures predicted by AlphaFold-2, a structure-prediction program, to AllergenAI, a machine learning allergen predictor. Model all proteins in the SDAP 2.0, COMPARE and AlgPred databases to improve the tool's ability to distinguish allergens from non-allergens.
- Test whether AllergenAI can identify which vicilins, a seed protein family that includes the peanut allergen Ara h 1 and its tree nut and legume relatives, cross-react with IgE antibodies in sera from allergic patients.
- Fine-tune AllergenAI's model settings using ESM-2, a protein language model, and benchmark the optimized tool against existing sequence-based allergen prediction methods.
- Optimize SDAP_AI, a convolutional neural network trained on allergen sequences from the SDAP 2.0 database, and test the robustness of its predictions on independent datasets.
- Use SDAP_AI to predict new pectate lyase allergens related to the ragweed allergen Amb a 1, along with their IgE-binding regions. Validate the predictions with peptide microarrays probed with sera from ragweed-sensitized patients.
- Re-evaluate previously reported IgE epitopes on Amb a 1, Jun a 1 and Hel a 6 with SDAP_AI and microarray tests. Map the confirmed linear epitopes onto allergen surfaces to define conformational epitopes, binding sites formed by the folded protein.
More about the research
Food allergies affect around 10% of people there, and peanut allergy is a leading cause of anaphylaxis, a severe whole-body reaction. Inhaled allergens such as ragweed pollen sensitize a large share of the US population. Patients must avoid triggers they often cannot see, and food makers must keep products free of unlabeled allergens. Both of this lab's projects use artificial intelligence (AI) to learn the sequence and 3D-shape features that make a protein allergenic. They also aim to predict which unstudied proteins could cross-react, meaning they could be recognized by the same IgE antibodies (the immune molecules behind allergic reactions). Both projects build on SDAP 2.0, the lab's Structural Database of Allergenic Proteins, and both test the computer predictions against blood serum from allergic patients.
Why it matters
Patients with food allergies currently manage them mainly by avoiding known triggers, and pollen allergies carry large clinical and societal costs. Tools that predict which proteins are likely allergens or cross-reactive could help clinicians and food manufacturers spot hidden allergy triggers. Defining the antibody-binding sites on allergens could also inform the design of better immunotherapies for food and inhalation allergies.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Stephen Dreskin
University of Colorado Denver/Anschutz Medical Campus, CO·Microbiology & Immunology
Optimize IgE binding to known 2S albumin peptides and discover new ones by positional amino acid screening, which systematically substitutes each residue in a peptide to find the strongest-binding version.
Stephanie Eisenbarth
Northwestern University, IL·Microbiology & Immunology
Test whether gut B cells must switch to IgG1 before switching to IgA to generate the highest-affinity IgA after oral immunization in mice, and map where this sequential switching occurs in mouse and human Peyer's patches using imaging and spatial transcriptomics.
Jennifer A Kim
Yale University, CT·Medicine & Clinical Research
Develop an algorithm that identifies post-traumatic epilepsy patients from electronic health records, and predict future PTE risk from the first seven days after TBI using structured data and natural language processing of clinical notes.
The school
University of Texas Medical Branch at Galveston: 14 labs on PI Pages, $241M a year on research, 141 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.