Independent Bioinformatics Researcher · Frisco, TX, USA
Naraen Palanikumar
Bioinformatics researcher working on immunology, single-cell and multi-omics data, and reproducible research.
Now Sole-author preprint, Runs that succeed and are wrong, auditing the validation runs of Bench, the multi-omics platform I built under Phalaena Automata. Google Cloud Professional Data Engineer.
Reading Oathbringer by Brandon Sanderson. More on the shelf.
About
Since July 2025 I have worked as an independent bioinformatics researcher. I reanalyze published datasets with the sample, not the cell, as the unit of inference (a 230,000-cell preleukemia reanalysis and a 140-donor tuberculosis TCR study), report what holds up, null results included, and build tools that make sequencing analyses reproducible. The largest is Bench, seven Nextflow pipelines for genomics, transcriptomics, proteomics and metabolomics data on Google Cloud, which I formed Phalaena Automata in March 2026 to host. In November 2026 I posted a preprint auditing Bench's own 42 validation runs: every run reported success, and 14 were wrong in ways its validation grading missed. Bench is now in maintenance mode.
Before that I was a research assistant at the Uniformed Services University, where I processed human specimens (PBMC isolation and culture, RNA and DNA extraction), immunostained cells and ran spectral flow cytometry, and analyzed the CD4 T cell arm of a published TCR repertoire study. Earlier, at the Sweeney Lab at UIUC, I built an interactive R Shiny tool, analyzed single-cell data, cryosectioned mouse brain samples and handled mice for colony upkeep, and worked clinical and emergency-response roles. Peer-reviewed contributions cover restricted TCR usage following SARS-CoV-2 vaccination (Frontiers in Immunology) and paraventricular thalamic MC3R circuits (The Journal of Neuroscience).
Browse the portfolio, read the CV, check out my gallery, or get in touch to collaborate.
Selected projects
Phalaena Automata
Bench
Python · FastAPI · React · Nextflow · Docker · Google Cloud Batch · Cloud Run · Vertex AI
Multi-omics analysis platform built as sole developer. Seven Nextflow pipelines for genomics, transcriptomics, proteomics and metabolomics run on Google Cloud Batch behind a FastAPI service on Cloud Run, with a results assistant (Vertex AI) that answers from each run's recorded summary. Now in maintenance mode and not taking new users. An audit of its 42 validation runs, posted as a preprint, found a confirmed silent failure in 14.
Preprint
Silent-failures audit
Python · pytest · JSON Schema · LaTeX
Sole-author audit of Bench's 42 validation runs: every run reported success, and 14 had a confirmed silent failure in five classes traced to specific code. The checks written afterward catch the targeted failures but were tested on only nine clean runs. The audit code, run registry and results are released as an open archive (Zenodo, doi:10.5281/zenodo.23717230).
Bioinformatics portfolio
Preleukemia and AML transcriptomics: single-cell + bulk RNA-seq
R · Python · Quarto · R Shiny · Seurat · Harmony · Scanpy
Two projects on the road into AML: a revised single-cell analysis of preleukemic mouse HSPCs (38 samples, eight mutation models) with a TCGA-LAML survival arm and an R Shiny dashboard, and a Nextflow bulk RNA-seq differential-expression pipeline with an interactive volcano.
Bioinformatics portfolio
TCR Clonality Analysis
Python · Jupyter · pandas · NumPy · SciPy
Bulk TCR-β repertoire analysis testing whether clonality and TB-specific clonotype frequency differ between TB progressors and controllers, revised to respect the cohort's matched case-control design.
Experience
-
Self-directed research (Phalaena Automata, formed Mar 2026) Jul 2025 - PresentIndependent Bioinformatics Researcher Frisco, TX (remote)
Reanalyzed a published single-cell dataset (230,000 bone marrow cells, 38 samples, eight preleukemic mutation models); re-tested the main comparisons across samples rather than individual cells, found that most of the original differences did not hold, and reported that null result. Compared T cell receptor repertoires of 140 donors from a tuberculosis cohort who did or did not progress to disease, using matched permutation tests and public reference sequences, and reported the null result plainly.
-
Uniformed Services University of the Health Sciences Oct 2024 - Jun 2025Research Assistant, Laboratory of Pediatric Infection and Immunity Bethesda, MD
Processed incoming human blood and nasopharyngeal specimens and ran high-parameter spectral flow cytometry (Cytek Aurora, FlowJo) alongside the bench team. Isolated PBMCs by Ficoll density gradient and cultured them, extracted RNA and DNA, and immunostained cells for flow cytometry in support of the lab's immunology studies.
-
Sweeney Lab, Molecular and Integrative Physiology, University of Illinois Aug 2021 - Jun 2023Undergraduate Research Assistant Urbana, IL
Built the lab's fiber photometry analysis application in R Shiny, replacing hours of per-session scripting; credited in the Journal of Neuroscience (2023) and still live online. Analyzed single-cell RNA-seq data to characterize MC3R-expressing neurons; cryosectioned mouse brain samples and handled mice for colony upkeep and documentation.