From watching a #Python tutorial with 13 yo son, a fun little #DataScience project has emerged: a script, that reads all SI files or full papers in a folder (e.g. the Download folder), extracts HRMS data and checks for internal consistency with the provided formulae. #ChemPython

Recently, I shared a demo of an HRMS checking tool for SI PDFs. I was asked whether reagents could be checked in a similar way. Yes, but in contrast to recalculating a chemical formula, obtaining the MW from the name of a reagent/chemical requires more information #ChemPython 1/2

Playing with data from 3000+ SI files. Searching for conditions and reagents, e.g. CDI, regarding scale, moles, solvent, yield, temperature, yield etc. If there are not too many outliers, M can be “crowdsourced” #ChemPython

Manually downloading ~2000 PDFs (5 GB) would be no fun. Instead, you can use a Python script to download all SI PDFs published in Beilstein JOC @BeilsteinInst from the last 20 years/volumes as training data set. #ChemPython

Thanks to everyone who shared their suggestions via DM or in the comments on expanding automated 🧪 SI checking. It’s evolving, thanks to #VibeCoding #ChemPython #DataScience

Impressed to see that the Accurate Mass (HRMS) Checker has already been used more than 1,000 times! Try it yourself: https://t.co/FyhUMNE88s #MassSpec #ChemPython #OpenScience #HRMS

Lab report grading got you down? 😩 We're building an automated feedback tool!  Share your pet peeves as regex in combination with a comment line. Example: r'\b\d+(?:.\d+)?\s+ml\b' catches "ml" vs "mL" #ChemEd #LabLife #AcademicTwitter #ChemPython

Well done, Gemini Pro 2.5: "Can you write a python script that generates a ball and stick model of a rotating water molecule that also show the two planes of symmetry and the C2 axis in a semitransparent fashion" #AI #GeminiAI #ChemPython #SciVis @GoogleAI

LLM-assisted tool generation is awesome. To extend the HRMS checker ( https://t.co/9saFLO7IWO....) to automated experimental checking, a #peptide sequence parser function (ca. 200 lines of code) was generated in under 2 hours #ChemPython #RealTimeChem #Coding

Catch mass (m) & substance amount (n) errors in PDFs!This tool comes handy, but expect false positives for now. 🔗 Link: https://t.co/y9dSiJ8Rsp Explore more #OpenChemTools: Accurate Mass: https://t.co/FyhUMNDAiU SI Check: https://t.co/l9MkQk8BPU #ChemPython #ChemTwitter

APIs like https://t.co/50JPnV0rZa or @pubchem as suggested by @MarcLucasAppis work fine, but they are the bottleneck. For the example above, it takes 224 seconds. Hence, looking for a faster way to get SMILES or MW from a name. Suggestion welcome #ChemPython

https://t.co/L2aplxu9pH