Open the cell model schema, then get stuck on this peptide

Someone has tried to treat biology as data engineering: collect enough omics, build a cell model, fill in the fields that should be filled in, and run it. rico meinl posted on X what he encountered after opening the schema; after reading it, you can roughly understand why this plan always gets stuck at the first molecule.

humanin, a peptide identified from brain tissue that remained viable after death in Alzheimer's patients. The sequence encoding it is hidden in the MT-RNR2 gene, which is normally considered the template for mitochondrial 16S rRNA. The same sequence translates to 24 amino acids in the cytoplasm and 21 in mitochondria. The reason is not complicated: mitochondria use a different genetic code, and the start amino acid is N-formylmethionine, while the cytoplasm uses ordinary methionine. The length question depends on which ribosome you ask.

That is not all. The nuclear genome still has 13 copies, each of which may produce its own version. This peptide is secreted into the blood and binds cell-surface receptors, one of which is itself a mitochondrial sensor. It is also taken back up into cells, where it binds Bax and blocks apoptosis. In other words, a single gene sequence is at once a structural component of the ribosome, a secreted signal, a hormone, and an anti-apoptotic function.

One reply put it precisely: for a peptide encoded within a ribosomal RNA gene, biology clearly has not heard of the single-responsibility principle.

Someone asked a very specific question: should the 24 aa and 21 aa forms be modeled as two separate entities, or as one molecule plus a translation-context field? This question is not easy to answer. If you choose two entities, do the 13 nuclear copies also each need one? If you choose one entity plus a field, then which version was measured in the receptor-binding experiment, and how do you backfill it? You can design the schema however you want; the molecule will not cooperate.

The original poster Tanishq Mathew Abraham's summary is: biology is messy and complex, nothing follows the rules, so it is very hard to 'solve.' Jacob Schreiber's reply takes another angle: how did they figure it out? Through data. That is true; humanin really was found by screening brain tissue, and no one first proposed a hypothesis and then went looking for it. But what the data gives is the outline of this molecule, not its classification. Aria Churchill pulls the conversation back: messiness and complexity are two different things; the problem is that people always want to abstract away complexity, stuff an entire class of behavior into a higher-level rule bucket, and then complain that the bucket cannot hold it.

The face in the video is still looping, captioned mfw, my face when. It is probably the expression shared by everyone trying to write a schema for cells. For compartment, fill in yes; for molecule_type, fill in rRNA, peptide, hormone; for the length field, fill in depends who's asking.