5 years ago I first saw 1 microgram of total peptide mammalian digest injected on column and produce almost 2,000 unique protein IDs (this was an Orbitrap Elite, HeLa digest, 2hr gradient). I still keep that RAW file on my desktop. It has given me a great benchmark in terms of peptide load and instrument performance. Since then I've seen more peptides/proteins ID'ed with much less material...but...it's gonna take a bit for me to wrap my head around...NANOGRAMs of peptides loaded, despite how inevitable it is that this will be commonplace soon.
This new ASAP study at JPR takes a look at how today's instruments (non ion-funneled quadrupole Orbitrap -- QE HF) can be optimized for these extremely low loads.
Some of the findings are obvious -- lower peptide concentrations are going to require higher ion injection times to get enough ions for good fragmentation. They end up using ion injection times over 250ms for their lowest quantity peptides.
One parameter I'm very curious to see for myself is widening the quadrupole isolation. At extremely low loads this group moves the quadrupole isolation to 4 Da -- nearly 3 times the 1.4 Da isolation width determined as ideal by researchers at the Max Planck Institute for this instrument! I've been using 1.4 Da exclusively for QE Plus and QE HF runs since that paper was released and I'm honestly at a loss here. I can't imagine any way widening this window does more than drive my FDR through the roof, because I'd be sampling a huge amount of space outside of the peptide isotopic envelope, but this is what these researchers report.
In the rationale, they state "...wider IW increases ion flux and ion transmission efficiency,[ref] which is crucial in trace proteomics..." I don't want to sound like a dummy, but there is a lot of maths in this paper that I don't have the qualifications to evaluate and this helps curtail my skepticism some. The peptide IDs reported seem to back these maths up.
There are other interesting deviations from the typical methodology reported in the paper -- for a complex mixture such as a mammalian digest, the consensus in the literature is generally that more resolution is better, yet this group uses an MS1 resolution of 60,000. A 15 cm column and 60 minute gradient is employed -- which probably makes sense for the low low peptide levels they are using, but for whole mammalian digests we generally like a lot more chromatographic resolution. The HCD collision energy utilized is NCE = 22, 5-8 units less than you generally see used.
I hope this doesn't seem critical -- I'm genuinely interested in what is going on here. A few months ago I might have said you were crazy if you thought that running 33 minute nanoLC gradients would provide the deepest human proteomes we've ever seen. Yet...the deepest proteomes I've ever seen were generated that way. It is worth noting that I did download this data and reprocess it the day after I read the paper... ;)
I would love to get these RAW files and take a look, but it doesn't appear they've been made publicly available yet. At the very least this is an interesting thought experiment. It is very easy to assume that the settings for one sample set may be directly applicable to the next without evaluation and this suggests I might be taking a few too many assumptions for granted.
No comments:
Post a Comment