Background

I’m Manu Arrojwala, a bioinformatics scientist interested in techbio AI roles and collaborations. I build computational methods and research systems for biological discovery.

My research experience includes chemoinformatics, behavioral genomics, and oncology drug discovery. In undergraduate research, I worked on metabolic modeling in Zymomonas mobilis and chemical-similarity and docking approaches for predicting protein targets.

At Georgia Tech, I combined computer vision and RNA-seq to study behavioral evolution in Malawi cichlids. That work was published in Scientific Reports in 2020.

Applied immuno-oncology

My work as a Bioinformatics Scientist at Arcus Biosciences began in February 2020. It spans scientific research, production bioinformatics, data architecture, and technical leadership across discovery, translational, and clinical oncology.

I have had primary bioinformatics responsibility for the AB801 AXL-inhibitor and Cbl-b-inhibitor programs, with contributions to AB598, etrumadenant/AB928, anti-TIGIT, and HPK1 programs. This account is limited to work I can discuss publicly and programs disclosed by Arcus.

Area My work
NGS and multi-omics Bulk RNA-seq, whole-exome sequencing, and single-cell RNA-seq: processing, quality control, analysis, and biological interpretation
Tumor-immune phenotyping Integrated transcriptomic and genomic features, including TMB, MSI, tumor content, neoantigen burden, and immune-state classifiers
T-cell repertoire modeling Diversity analysis, ensemble treatment-response models, and exploration of latent repertoire representations
Neoantigens and TCR specificity HLA typing, tumor/normal variant calling, peptide–MHC prediction, immunogenicity ranking, and candidate TCR–neoantigen matching

This work includes target discovery and validation, preclinical mechanism-of-action studies, clinical biomarkers, and patient stratification.

Data platforms and pipelines

I designed a centralized multi-omics platform for multiple clinical trials, including databases and a clinical data mart. This included reorganizing clinical, translational, and biomarker data across cloud and on-premises storage with appropriate access controls.

I replaced legacy bcbio workflows with production Nextflow pipelines for RNA-seq and WES and led compute migrations between Slurm and AWS Batch. The work improved throughput and cost per sample; proprietary operational metrics are not reproduced here.

I also built internal scientific applications using Shiny, FastAPI, PostgreSQL, and R for dataset discovery, processing, and curated gene sets. Roughly twenty scientists used these applications daily, replacing manual discovery and curation.

Technical leadership

I formally managed three interns over time across engineering, bioinformatics, and ML research: project scoping, assignment, mentoring, review, and performance evaluation.

Separately, I served as technical lead for two peer engineers, covering architecture, planning, code review, mentoring, and delivery coordination. I led cross-team design discussions for data platforms, pipelines, storage, and clinical-data organization.

Scientific AI research

My oncology work includes data acquisition, cohort definition, preprocessing, batch-effect analysis, provenance tracking, and biological interpretation.

I apply these methods in biology-agent evaluation, hypothesis generation, and method and evidence checks. I examine the source data, transformations, intermediate artifacts, and assumptions behind agent outputs.

Publications and abstracts

Collaborations and roles

I’m interested in techbio AI roles and collaborations in scientific-agent systems, evaluation, research engineering, and reproducible biological analysis.

Email me or connect on LinkedIn.