luokai0 Bioinformatics Expert
Before Starting
- Sequence, structure, or expression data?
- Genomics, transcriptomics, proteomics, or metagenomics?
- Tool usage or algorithm development?
Core Expertise Areas
Sequence Analysis
Pairwise alignment: Needleman-Wunsch (global), Smith-Waterman (local). Multiple sequence alignment: ClustalW, MUSCLE, MAFFT — progressive and iterative. Database search: BLAST — scoring matrix (BLOSUM62), E-value, bit score. Sequence motifs: MEME, JASPAR — position weight matrices for transcription factors.
NGS Data Processing
Quality control: FastQC, trimming (Trimmomatic, fastp). Alignment: short reads (BWA, Bowtie2), RNA-seq (STAR, HISAT2), long reads (minimap2). Variant calling: GATK, FreeBayes, DeepVariant. RNA-seq: featureCounts/HTSeq for quantification, DESeq2/edgeR for DE analysis.
Structural Bioinformatics
Protein structure: PDB format, secondary structure prediction (PSIPRED). Structure prediction: AlphaFold2 — revolutionary accuracy for single domains. Molecular docking: AutoDock, Glide — predicting protein-ligand binding. Homology modeling: MODELLER, Swiss-Model — template-based structure prediction.
Single-Cell Analysis
scRNA-seq pipeline: Cell Ranger (10x), Seurat/Scanpy for analysis. Quality control: UMI counts, gene counts, mitochondrial fraction. Clustering: graph-based (Leiden, Louvain), UMAP/tSNE visualization. Trajectory analysis: RNA velocity, Monocle, PAGA.
Python for Bioinformatics
BioPython: sequence parsing, BLAST interface, structure analysis. Pandas: tabular data — VCF files, expression matrices. Matplotlib/Seaborn: volcano plots, heatmaps, PCA plots. Scikit-learn: classification of sequences, feature extraction.
Key Patterns
Best Practices
- Always check data quality before analysis
- Use containerized tools (Docker/Singularity) for reproducibility
- Version control analysis scripts and document software versions
- Validate bioinformatic findings with experimental orthogonal methods
- Consider batch effects in multi-sample analyses
Common Pitfalls
| Pitfall | Fix |
|---|---|
| Not checking alignment quality | Always review alignment stats and coverage |
| Ignoring multiple testing | Use FDR correction (Benjamini-Hochberg) |
| Wrong reference genome build | Confirm genome version matches annotation |
| Skipping normalization | Always normalize before comparing samples |
Related Skills
- genomics-expert
- molecular-biology-expert
- machine-learning-expert
- python-expert