<aside> 🧬
I made this page for one simple reason. I have spent a lot of time working with real biological data, but that is not always obvious from a resume or a LinkedIn profile.
So this is my little data scrapbook. A place to show the kinds of biological datasets I have actually worked with, from genome structure and RNA sequencing to proteomics, protein embeddings, and 3D protein structures.
</aside>
Genome and chromatin
Micro C Hi C MNase seq ATAC seq
RNA
Short read RNA seq Oxford Nanopore RNA seq Transposable elements
Proteins
Mass spectrometry Phosphoproteomics Acetylomics Glycoproteomics
Protein AI
ESM 2 embeddings UniProt sequences AlphaFold structures
This is where I got comfortable working with very large sequencing datasets and learning how genome organization changes depending on the assay, the cell line, and the biological question.
| Dataset | Data type | What it contains | Scale or context | Source |
|---|---|---|---|---|
| GSE68016 | Micro C | S. cerevisiae 3D genome contact data | 20 replicates, about 74M pooled interactions | GEO |
| GSE53721 | MNase seq | Yeast nucleosome positioning data | Used alongside 3D genome analysis | GEO |
| ENCSR000CXP | MNase seq | GM12878 nucleosome positioning data | Human chromatin data | ENCODE |
| ENCSR000CXQ | MNase seq | K562 nucleosome positioning data | Human chromatin data | ENCODE |
| GSE44985 | MNase seq | IMR90 chromatin and nucleosome data | About 697M reads | GEO |
| HCT116 · GSM3889219 | MNase seq | HCT116 colorectal carcinoma nucleosome positioning data | Paired end MNase seq | GEO |
| HepG2 · GSM3718063 | MNase seq | HepG2 nucleosome positioning data | 176.4M paired end sequencing spots | NCBI SRA |
| GM12878 · ENCFF053VBX | Hi C | 3D chromatin contact data | GM12878 | ENCODE |
| K562 · ENCFF080DPJ | Hi C | 3D chromatin contact data | K562 | ENCODE |
| IMR90 · ENCFF685BLG | Hi C | 3D chromatin contact data | IMR90 | ENCODE |
| HCT116 · ENCFF750AOC | Hi C | 3D chromatin contact data | HCT116 | ENCODE |
| HepG2 · ENCFF020DPP | Hi C | 3D chromatin contact data | HepG2 | ENCODE |
| GM12878 · ENCFF415FEC | ATAC seq | Open chromatin data | GM12878 | ENCODE |
| K562 · ENCFF512VEZ | ATAC seq | Open chromatin data | K562 | ENCODE |
| IMR90 · ENCFF848XMR | ATAC seq | Open chromatin data | IMR90 | ENCODE |
| HCT116 · ENCFF724QHH | ATAC seq | Open chromatin data | HCT116 | ENCODE |
| HepG2 · ENCFF239RGZ | ATAC seq | Open chromatin data | HepG2 | ENCODE |
Tools I used: Bowtie2 · Samtools · Cooler · hicstraw · pyBigWig · DANPOS · NucPosSimulator · Python · SLURM
Proteomics is probably the area where I have spent the most time dealing with messy biological evidence, conflicting observations, and the gap between what an experiment measured and what a model eventually sees.
| Dataset | Data type | What it contains | Context | Source |
|---|---|---|---|---|
| PXD005793 | Acetylomics and proteomics | Lysine acetylation mass spectrometry data | PTM site data | ProteomeXchange |
| MSV000082644 | Proteomics and multiomics | Medulloblastoma proteome, acetylome, and phosphoproteome data | Cancer multiomics study | MassIVE |
| MSV000078509 | PTM proteomics | Acetylation and phosphorylation measurements | Integrated PTM data | MassIVE |
| PXD012174 | Phosphoproteomics | Large scale human phosphorylation data | Serine, threonine, and tyrosine sites | ProteomeXchange |
Tools I used: Python · Pandas · NumPy · UniProt · FASTA · PyTorch · Linux · SLURM · HPC
I also worked with outputs from several glycoproteomics tools, which was useful because each tool structures peptide, glycan, protein, and scan information a little differently.
| Dataset or tool output | Data type | What it contains |
|---|---|---|
| StrucGP | Glycoproteomics | Glycopeptides, glycans, proteins, glycosites, scans, and MS file information |
| pGlyco3 | Glycoproteomics | High confidence glycopeptide identifications |
| GlycoDecipher | Glycoproteomics | Glycan and glycopeptide identification outputs |
| MSFragger | Proteomics and glycoproteomics | Mass spectrometry search results used alongside glycoproteomics outputs |
This work pushed me into a very different kind of biological ambiguity. Instead of asking which peptide maps to which protein, the problem becomes which genomic locus a sequencing read actually came from.