RNA Biology x Computational Genomics x AI for Science

Yuchen Cai

Ph.D. researcher at the State Key Laboratory of Genetic Engineering, Fudan University, building biology-aware computational methods, reproducible omics workflows, and evaluation systems for biological AI.

Computational frameworks for RNA regulation.

My work connects sequencing evidence, genetic mutants, machine learning, and biological model evaluation. I care about methods that are not only accurate, but also inspectable, reusable, and useful to experimental teams.

2019-2026

Ph.D. Researcher, Fudan University

Developed small RNA analysis frameworks, miRNA classifiers, mutant-informed annotation strategies, and RNA-end modification workflows.

Now

Foundation models for biology

Exploring RNA language models, multimodal biological representations, and benchmarks across DNA, RNA, protein, and omics data.

Research questions turned into usable systems

Research pipeline

Integrated plant small RNA-seq analysis

TRMRNAseqTools

Problem

Plant small RNA studies often depend on fragmented preprocessing, annotation, and counting steps that are difficult to reproduce across datasets.

My contribution

I built a Shell and Python workflow that automates read processing, hierarchical RNA annotation, length profiling, and miRNA re-annotation.

Value

A reusable analysis path from raw small RNA-seq reads to interpretable plant RNA summaries.

Plant small RNAPipelinePython
View repository

Method development

Parallel profiling of miRNA 3-prime modifications

Tailing / trimming workflow

Problem

Non-templated tailing and trimming events are biologically informative but difficult to recover and compare across many small RNA-seq samples.

My contribution

I consolidated the analysis into a parallel workflow that reconstructs modification profiles, supports UMI libraries, and integrates genome and feature-level summaries.

Value

Restartable multi-sample analysis with per-miRNA profiles, tail-base statistics, and publication-ready visual outputs.

miRNARNA modificationReproducibility
View repository

Published research

Reproducible analysis for RNA modification discovery

PBOX-sRNA-seq

Problem

Specialized small RNA libraries need analysis code that connects sequencing evidence to specific terminal RNA modifications.

My contribution

I contributed the analysis code used to generate results for the PBOX-sRNA-seq study published in Nucleic Acids Research in 2024.

Value

A transparent computational record supporting discoveries in miRNA modification and modified 5-prime tRNA fragments.

PublicationsRNA-seqRNA modification
View repository

Research engineering

From raw reads to differential and splicing analysis

Bulk RNA-seq workflow

Problem

Bulk RNA-seq projects become fragile when quality control, mapping, counting, differential expression, and reporting are run as disconnected scripts.

My contribution

I designed a modular 13-step workflow covering validation, QC, rRNA filtering, mapping, gene and miRNA counting, DE analysis, alternative splicing, and reporting.

Value

A restartable and auditable workflow that keeps analysis definitions, contrasts, and outputs aligned across a study.

RNA-seqDESeq2Workflow
View repository

Research engineering

One-command methylome processing and reporting

WGBS analysis

Problem

WGBS analysis combines expensive alignment, context-aware methylation calls, cohort QC, and many downstream figures that must remain restartable.

My contribution

I built a validated entry point for QC, Bismark mapping, methylation extraction, metaprofiles, summaries, and publication figures.

Value

A reproducible methylome workflow with configuration checks that prevent incompatible results from being silently reused.

EpigenomicsWGBSAutomation
View repository

Visualization tool

Inspect small RNA evidence in genomic context

sRNA Track Viewer

Problem

Small RNA loci are easier to evaluate when read distributions and annotations can be inspected together instead of compared across static files.

My contribution

I developed a focused viewer for exploring small RNA sequencing tracks and locus-level evidence.

Value

A more direct way to review patterns, compare regions, and communicate sequencing evidence.

VisualizationSmall RNAGenomics
View repository

What my work is designed to answer

How can genetic evidence improve plant miRNA annotation?

I combine mutant-aware sequencing signals, hierarchical annotation, and machine-learning classifiers to distinguish high-confidence miRNAs from ambiguous small RNA loci.

How should biological foundation models be evaluated?

I focus on biologically meaningful tasks, reproducible evaluation pipelines, and comparisons that connect learned representations to RNA, DNA, protein, and omics outcomes.

How can omics workflows become easier to trust?

I turn analysis decisions into restartable pipelines with validation, explicit configuration, consistent reports, and outputs that can be traced back to the source data.

Useful at the boundary of science and engineering

Biology-aware computation

I translate questions in RNA regulation into analysis definitions that remain grounded in experimental evidence and biological interpretation.

Research-grade engineering

I build workflows that move beyond one-off scripts: modular, restartable, reviewable, and practical for real multi-sample studies.

AI evaluation with scientific context

I connect foundation-model methods with the domain knowledge needed to choose meaningful tasks, controls, and evidence standards.