About Me

IMG_1204Welcome! I am a Ph.D. candidate in the MCB (Molecular and Cellular Biology) program at Dartmouth College, working with Prof. Gevorg Grigoryan (lab page). I expect to finish my degree in the fall of 2016.

My research is about computational structural biology. Understanding the protein structures is critical for studying natural biological processes, drug discovery and protein engineering. Our lab is especially interested in protein design, which is to engineer the amino-acid sequences of proteins to make them optimal in folding into desired 3D structures. My projects are mainly on two directions:

1. Data mining of protein structures. A convenient approach to design a protein is to learn from the nature. The PDB database for experimentally solved protein structures has been an important resource to learn protein sequence-structure relationships and extract rules for protein design. The statistical rules of protein secondary structures have been well established. However, the traditional rules have been insufficient when scientists are getting more ambitious and want to design more complicated geometries. A major theme of my research is to study the sequence-structure relationships on the level of tertiary motifs (TERMs), which are structural segments associated with inter-residue contacts, being more complicated than secondary structures. I have shown that useful statistical propensities can be extracted from the statistics of TERMs, and I developed methods that convert the statistics into evaluating scores, which have been used to improve the prediction of protein structures and the protein stability upon amino acid mutations.

2. Specific peptide design to target protein-protein interactions. Proteins work in complexes, and studying their interactions is critical to understand their functions and develop therapeutics. Beginning with protein-peptide interactions, which account for about 40% of all protein interactions in cells, I combined peptide docking, molecular dynamics and machine learning methods to accelerate the computational assessment of protein-peptide affinities for PDZ domains. The speed-up is critical in design of specificity, when the designed peptides are required to bind the target without significant cross-reactivity with structurally similar domains. Successfully designed peptides have been experimentally characterized, and our collaborators have shown potential in vivo functions of these peptides.

For details and related publications, check out the “Research” page.

A few facts of me:

  • A few Python packages/tools I like: iPython, Numpy, Jupyter Notebook, ProDy
  • A few R packages/tools I like: Markdown, ggplot2, dplyr, Shiny
  • My favorite IDE: PyCharm
  • My favorite website: Coursera

Get every new post delivered to your Inbox.