Projects
A selection of projects I’ve worked on.
RNA-seq Pipeline for Rare Diseases
A comprehensive Nextflow-based RNA-seq analysis pipeline built for the UCLA Nelson Lab, designed to run on AWS HealthOmics. Processes paired-end RNA-seq data through quality control (Fastp), contamination detection (rRNA/Globin via BWA), splice-aware alignment (STAR with multiple read-length indices), duplicate marking (Sambamba), gene and transcript quantification (FeatureCounts + Kallisto), intron retention detection (IRFinder), coverage analysis (Mosdepth), variant calling on well-covered coding regions (DeepVariant with custom RNA-seq model), and splice junction analysis (BAM2SJ) — all containerized with Docker and orchestrated through AWS ECR. Built to support the Undiagnosed Diseases Network (UDN) and rare disease diagnostics.
Nextflow, Docker, AWS HealthOmics, STAR, DeepVariant, Kallisto, IRFinder, Samtools
WGS Pipeline for Rare Diseases
A Nextflow-based whole genome sequencing pipeline for germline short-read data, built for the UCLA Nelson Lab and designed to run on AWS HealthOmics. Supports FASTQ, BAM, and CRAM inputs with flexible subworkflows. Includes alignment (BWA-MEM2), duplicate marking (Picard), SNV/indel calling (DeepVariant), structural variant detection (Manta), copy number analysis (CNVpytor), repeat expansion detection (ExpansionHunter + ExpansionHunterDenovo), haplotype phasing (HapCUT2), runs of homozygosity (AutoMap), and comprehensive QC (Fastp, Qualimap, Picard metrics, MultiQC). All tools containerized with Docker and orchestrated through AWS ECR.
Nextflow, Docker, AWS HealthOmics, BWA-MEM2, DeepVariant, Manta, HapCUT2, CNVpytor
Aneuploidy Analysis from WGS
Genome-wide aneuploidy assessment from whole genome sequencing data using AStra and QDNAseq. Performed as part of a published case study in Haematologica investigating fraternal twins with HJV mutations and discordant hemochromatosis severity. The analysis ruled out chromosomal abnormalities in a trio (proband + parents) by generating aneuploidy profiles, spectra, and copy number calls from standard-depth WGS.
AStra, QDNAseq, R, WGS, Copy Number Analysis
Simple Text-to-Speech
A command-line tool for converting text files to MP3 audio using Microsoft Edge TTS neural voices. Supports customizable speech rate, multi-language voice selection, progress tracking, and simple CLI interface. Useful for creating audiobook-style recordings from plain text files, accessibility, or language learning.
Python, edge-tts, Microsoft Edge TTS, CLI
Single-Cell RNA-seq Pipeline
A Nextflow pipeline for processing 10x Genomics single-cell/single-nucleus RNA-seq data, built for the UCLA Nelson Lab and designed to run on AWS HealthOmics. Wraps Cell Ranger for gene expression quantification and cell calling, with FastP for QC, automatic BAM-to-CRAM conversion for 40-60% storage savings on S3, and MultiQC for aggregated reporting. Supports configurable chemistry detection, adjustable expected cell counts, and optional alignment file generation — all containerized with Docker via AWS ECR.
Nextflow, Docker, AWS HealthOmics, Cell Ranger, Samtools, FastP, MultiQC