geohot/corona

Reverse engineering SARS-CoV-2


repo name	geohot/corona
repo link	https://github.com/geohot/corona
homepage
language	Python
size (curr.)	475 kB
stars (curr.)	655
created	2020-03-23
license

Reverse engineering the Coronavirus

How is orf1ab cleaved into polypeptides? Can we predict this from the sequence?
How do the researchers know (guess?) where orf1ab cleaves?
- nsp3 and nsp5 do it – https://www.pnas.org/content/pnas/103/15/5717.full.pdf
Which protein is the immune system responding to?
- “spike” and “nucleocapsid” – http://www.cmi.ustc.edu.cn/1/3/193.pdf
- Are some people already immune from exposure to other coronavirus?
Find the “furin cleavage site” in the “spike glycoprotein”
- It might be at the “PRRA” – https://www.sciencedirect.com/science/article/pii/S0166354220300528
- Use ProP or PiTou to predict? – https://en.wikipedia.org/wiki/Furin
How similar are the other coronaviruses? (causes colds, not either SARS or MERS)
- alpha
  - https://en.wikipedia.org/wiki/Human_coronavirus_229E (simpler, though targets APN)
  - https://en.wikipedia.org/wiki/Human_coronavirus_NL63 (targets ACE2!)
    - https://www.ncbi.nlm.nih.gov/nuccore/MG772808
- beta
  - https://en.wikipedia.org/wiki/Human_coronavirus_OC43 (targets Neu5Ac)
    - https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2095096/pdf/JIDMM17330.pdf
    - Specifically, how similar is the N protein OC43, SARS v1, and SARS v2?
  - https://en.wikipedia.org/wiki/Human_coronavirus_HKU1 (targets Neu5Ac)
  - MERS-CoV
  - SARS-CoV
  - SARS-CoV-2
What adds the phosphate group to the N protein? Kinase?

Each letter in genome is a “byte at address x”
Translation = Disassembly, it’s a 3 byte wide instruction set with arbitrary “reading frames”
Protein ~ Function. polyprotein = Function with multiple pieces
Proteins appear to have “basic blocks” = “Secondary Structure”
- 80% accuracy in prediction: https://en.wikipedia.org/wiki/Protein_structure_prediction#Secondary_structure
“Tertiary Structure” forms a function
- This seems like the hard one to predict: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0205819
“Quaternary Structure” is like compiled function with inlining
- https://en.wikipedia.org/wiki/Protein%E2%80%93protein_interaction_prediction
Gene ~ library (bacteria are static linked, viruses are dynamically linked)
Transcription = loading off disk

There is no equivalent to execution, we are reverse engineering a CAD format
Static analysis, looking at the DNA, protein structure prediction, FLIRT signatures, etc…
Simulation doesn’t seem to work yet
Tons of in system dynamic analysis, but the tools are crap
Runs more like FPGA code, all at once, no serial execution (what are the FPGA re tools?)