Brian Wiley

I have been working in Bioinformatics Research and Structural Biology and Computer Aided Drug Design for last 2 years at the University of Washington in St. Louis.

Currently I am working with Dr. Jason Held at Washington University, one of my references, on a project that we are very early on in, to decrypt solvent accessible cysteine pockets in cancer associated proteins with PDB structures to look at potentially drugging cysteines, i.e. EGFR-afatinib is an example where this has been successful.  We are using a software that implements Markov state models to hypothesize whether cysteines become solvent accessible only after post-translation modifications, in the case for Erk2, the first protein we are testing on, being dually phosphorylated at Thr183 and Tyr185.  It was actually a rather tough job to get the adaptive sampling python wrapper workflow that was designed by a PhD student to analyze and interactively run molecular dynamics simulations on our IBM LSF Spectrum HPC.  Not only did we require installing the GROMACs software to run with GPU and though MPI threading, but also had to create multiple docker images and changes to the software to get it to run and run efficiently.  We learned the hard way that running multiple GROMACs simulations on the same GPU device would slow down the multiple simulations astronomically and unfortunately the IBM’s scheduler fills up single devices before moving onto others but I found a simple fix. Additionally we will implement Shake-Rupley analysis for scoring levels of changes in SASA around these cysteines. What we have learning so far in running our first experiment on phosphorylated Mapk1/Erk2 is that we needed to run longer simulations and more of them to sample all the conformation space, and we still have not seen the inactive D-outFG motif flip to active D-inFG motif despite phosphorylation as we have not included any ATP or Mg2+. Additionally I ran some molecular dynamics with Tinker on hypothesized 3-water molecule coordinated crystal structure 4JPS for PI3KCA-Alpelisib but interestingly only 1 of the 3 water molecules remained in the binding pocket at end of the simulation.  Working with coordinated water-molecule and drug binding affinity is exciting and vastly important but further exploratory work and analysis by myself and others would be needed however for this structure. Another topic I am interested in and have run simulations on so far are for an ABL1 structure with both dasatinib add first and inducing imatinib to bind allosterically which can actually activate ABL1.  This is interesting because imatinib resistant mutation in the active site, such as G250E mentioned above, would induce this allosteric binding and activation. The benefit for this work would help find therapeutics in patients with CML which must take imatinib indefinitely if or until they become imatinib resistant.  I believe working with researcher sat UNC would help me in my lead discovery of replacement compounds that bind imatinib-resistant mutant ABL1. 

For my current research as a Bioinformatics Scientist at Washington University in St. Louis I running variant calling in blood samples to detect mutations and mutational signatures from general healthy populations as well as pediatric and adult leukemias.  The tools I have mostly using are workflow languages/interfaces like CWL, WDL and DNAnexus bash pipelining for the UKBioBank.  We are using CLI variants callers such at GATK’s Mutect2 and HaplotypeCaller, VarDict, VarScan, etc.  Most of my downstream and analyses are done with R and calling API’s from both R and Python for variant annotation as well writing bash scripts to run every day analysis and jobs on our IBM LSF compute HPC.

JOHNS HOPKINS UNIVERSITY, Baltimore, MD

Master’s of Science in Bioinformatics                                 June 2019 – August 2021        Major GPA:   3.9