run a blank
#hplc-massspec 2025-01-25
Saturday26 messages6 participantstimes are UTC
Highlights from this day
- snac_and_water — while im here if the number you are staring at is a third of what you expected, do the charge state arithmetic before you panic 12:38
- blank_run — a dimer usually shows up late and at roughly double the mass. the chromatogram alone will not tell you 14:06
- area_percent — semaglutide is about 4113.6 Da. [M+2H]2+ lands near 2057.8, [M+3H]3+ near 1372.2 16:08
co-elution is the failure mode that a beautiful single peak is most likely to be hiding
while im here if the number you are staring at is a third of what you expected, do the charge state arithmetic before you panic
⚠️8❤️2📈3
within inter-lab range
does retention time on its own tell you what the compound is
shallow gradients hide related substances behind the main peak. a steeper one separates and looks worse, honestly, thats one data point
retention time is a hint
a dimer usually shows up late and at roughly double the mass. the chromatogram alone will not tell you
🧊1🙏3📉6
214nm sees the amide backbone so it sees everything. 280 only sees aromatics
the first peak is the solvent. it is always the solvent
relative retention time against a standard is comparable between labs. absolute minutes are not, we shall see
does ms/ms actually prove identity or just make it very likely
i was wrong about tailing factor. it is usually the injection or the pH, not the column
update on the earlier thing why do two labs integrate the same peak differently
plus 0.98 is deamidation
semaglutide is about 4113.6 Da. [M+2H]2+ lands near 2057.8, [M+3H]3+ near 1372.2
🤝10🔥2✅11
retention time is a hint, not a fingerprint. two different peptides can co-elute perfectly happily
truncations show as a mass deficit matching one or more residues. you need the sequence to say which
how do you tell co-elution from a clean single peak
is a tailing factor of anything specific a problem
baseline drift on a gradient is normal. baseline drift on an isocratic hold is your column
a shoulder on a peak is information, and usually bad information
you need more sample for identity than for purity. plan for that when you post it
i asked VendorInvestigate for the raw trace and they sent it. that is the right answer to that question
[edited]nobody runs enough blanks, including me
[edited]integration choices move the number by a few tenths easily. that is most of your inter-lab variance