Sightline: the cysteines your experiment was never going to see

Chemoproteomics reports what it detected. An undetected cysteine isn't necessarily unreactive. It may sit on a peptide too long or too short to identify, too hydrophobic to elute, or one that never ionizes well. Everyone in the field knows this. Sightline turns it into a tool.

01What it returns

Give it a target and the workflow you plan to run (probe, protease, cell background). For each cysteine it returns:

• Observable by this workflow: yes, no, or marginal.
• If not, why not, broken into named causes rather than one score: no peptide of identifiable length after this protease, too hydrophobic, poor predicted ionization, protein not expressed in that background.
• The single change that fixes it. Usually a protease swap, sometimes a different cell line. Sometimes the answer is nothing, and that's useful too: the site needs a different kind of experiment.
• Accessibility, reported separately. "The chemistry can't reach it" and "the instrument can't see it" are different claims, and Sightline keeps them in different columns.

The deliverable a wet lab actually uses is a protocol card: for the site you care about, the smallest workflow change that sees it.

02Why it matters for every other model

Any reactivity model trained on detected versus undetected sites is partly a model of the mass spectrometer. Sightline is that model, built on purpose and pointed at experimental design rather than tucked into a limitations paragraph. It grew out of the detectability baseline that my hackathon team's project, DarkSite, had to beat.

03Built with

Python: in-silico digests across proteases, peptide identifiability and hydrophobicity rules, and per-experiment detection history.

04Status

A side project from October 2026. The code is private because it runs on unpublished research data.