Mark Zhai

I conduct computational protein engineering research, combining AI-driven protein design and molecular dynamics simulations to develop novel molecular recognition tools and optimized CRISPR-Cas12a systems.

About

Hi! I am a chemical engineering and computer science graduate, and current research technician in a molecular diagnostics lab at Tulane University School of Medicine. There I lead two computational protein design projects and assist with data analysis for others.

My goal is to leverage the advantages of both computational and experimental techniques to answer my research questions.

Current Research

Generative design of Cas12a nucleases

Using reinforcement learning to fine-tune a protein language model on the Cas12a family to generate novel nucleases with improved thermostability, specificity, and activity to optimize one-pot CRISPR detection assays.

ProGen2 · AMBER · sequence level filtering · OpenRLHF

Computational nanobody design for difficult targets

Expanding the scope of known computational binder design pipelines to small tryptic and disulfide-rich peptides with improved specificity for detection and neutralization purposes.

RFdiffusion · AlphaFold2 Sequence Hallucination · ProteinMPNN · Rosetta ddG

Transcriptomics of host–pathogen signaling

Bulk RNA-seq analysis supporting the lab's infection biology work, including alignment, quality control, and differential expression pipeline.

STAR · FastQC/MultiQC · featureCounts · DESeq2

Previous Research

Downregulated pathways involved in ZFTA-RELA ependymoma tumorigenesis

Identified key downregulated pathways implicated in ZFTA-RELA ependymoma tumorigenesis, corroborated by their presence in in-vitro and in-vivo models.

bulk RNA-seq analysis · ChIP-seq analysis · R

Automated MRI tumor segmentation for fast tumor volume quantification

Built and trained a U-Net convolutional neural network to automate unbiased MRI tumor segmentation for fast tumor volume quantification.

PyTorch · CNN · Slicer · Python

IGF2 secretion as a mechanism of ZFTA-RELA ependymoma tumorigenesis

Constructed growth curves for multiple patient-derived xenograft ependymoma models to measure the effects of IGF2 treatment.

Cell culture · Western blot · ELISA · PCR · RNA extraction

Publications

  1. Ippagunta, S. M., Arabzade, A., Varadharajan, S., ..., Zhai, M., ..., Mack, S. C., & Bertrand, K. C. Nuclear export as a therapeutic vulnerability in ZFTA-RELA ependymoma. Neuro-Oncology, 2026. DOI
  2. Arabzade, A., Shirnecki, H. K., Varadharajan, S., Ippagunta, S. M., ..., Zhai, M., ..., Kriwacki, R. W., & Mack, S. C. Synthetic ZFTA fusions pinpoint disordered protein domain acquisition as a mechanism of brain tumorigenesis. Nat Cell Bio, 2025. DOI

Conference Abstracts & Presentations

  1. Zhai, M., Kim, Y., Fan, J., Roy, C., & Ning, B. De novo binder design for detection of small peptide analytes. American Chemical Society Spring 2027 National Meeting., 2027. Abstract submitted
  2. Zhai, M., Syamsudin, S., & Ning, B. Structure- and function-guided fine-tuning of a protein language model for engineering Cas12a effectors. American Chemical Society Spring 2027 National Meeting., 2027. Abstract submitted

Research Experience

  1. 2026 — present Research Technician IIITulane University School of Medicine
  2. 2025 Bioinformatics InternSt. Jude Children's Research Hospital, Department of Developmental Neurobiology
  3. 2024 — 2025 Undergraduate ResearcherTulane University School of Science and Engineering
  4. 2024 Research InternSt. Jude Children's Research Hospital, Department of Developmental Neurobiology
  5. 2023 Pediatric Oncology Research InternSt. Jude Children's Research Hospital, Department of Developmental Neurobiology

Education

  1. 2022 — 2026 B.S.E Chemical Engineering & Computer ScienceTulane UniversityMagna Cum Laude