Mengyao Zhao
Software Algorithm Team Manager (Staff Computational Biologist) @Pillar Biosciences Incorporated
Signup · Get unlimited contacts
WORK HISTORY
Software Algorithm Team Manager (Staff Computational Biologist) @Pillar Biosciences Incorporated
Natick, MA, US
Led the Software Algorithm Team and outsource contractors in maintaining and enhancing Pillar\'s amplicon sequencing-based somatic Variant Analysis Toolkit (PiVAT) for In Vitro Diagnostics (IVD) and Research Use Only (RUO) pipelines. PiVAT detects various somatic variants, including SNPs, short and long InDels, CNVs, fusions, MSI and methylations. We implemented four major improvements:1. Optimized PiVAT\'s variant calling pre-processing modules by rewriting them in pure C. Together with algorithmic optimization, we achieved ~24-fold speed improvement and ~99.9% memory usagereduction (from ~59G to ~66M) compared to the previous Python/Cython equivalent modules. For the entire PiVAT pipeline, this improvement achieved 49.6% run time reduction, and 49.32% memory usage reduction.2. Improved our local realignment and long InDel, including internal tandem duplication (ITD), callingalgorithm. This algorithm change dramatically improved FLT3 gene ITD region InDel calling accuracy.3. Restructured PiVAT into separate sub pipelines. Separated the data analysis pipeline codes from the user platform codes. This architectural enhancement enabled flexible pipeline usage and independent platform and pipeline development cycles.4. Developed new pipelines to support bladder cancer treatment clinical trials, and HDFN and FNAIT treatment CDx in collaboration with Johnson and Johnson.(C, Python, Cython, unit testing, module testing, regression testing, Bamboo CI/CD, Docker)• Maintained the Stripped Smith-Waterman (SSW) Library for both internal usage and public releases. The SSW library is ~50 times faster than ordinary Smith Waterman implementations, and has been used in Pillar\'s PiVAT, Illumina\'s Dragon, Google Deepvariant, and Nvidia\'s Parabricks Deepvariant.• Offered PiVAT-related customer support to ensure smooth operation and address user inquiries.
EDUCATION
The University of Edinburgh
Master of Science (MSc), Genomics and Pathway Biology
Boston College
Doctor of Philosophy (PhD), Bioinformatics
Beijing Normal University
Bachelor of Science (BSc), Computer Science
ABOUT MENGYAO ZHAO
Twenty years of programming experience in bioinformatics. Eighteen years of large genomic data analysis experience. Independently designed novel algorithms including dynamic programming and parallel computing to solve next generation sequencing data analysis problems. Led software development and keep active software maintenance.Specialties: C/C++, Python (NumPy, SciPy, pandas), Perl, SQL, R, Unix/Linux, cluster usage (LSF/SGE/Slurm), high performance computing, machine learning (TensorFlow/Keras, CNN/MLP/HMM), large-scale NGS data analysis and pipeline development (Snakemake), cancer genomic (DNA/RNA) data analysis, cloud computing (AWS/Google cloud, Docker), HLA typing, population/evolutionary genetics, system biology, biomarker development, CDx development, clinical trial, AV writing, FDA filing
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.