Monday, December 17, 2007

Three kinds of display

M provided a link to a page at the American Society for Mass Spectrometry that included the following very useful illustration identifying the type of displays that can be provided.

We see identified three kinds of traces.

  1. Total Ion Count (TIC) reflecting the counts of ions of all masses for the entire run.
  2. Selected Ion being monitored (SIM) the count of particular ions at selected masses for the whole run
  3. Mass spectra for individual peaks, in this case we think a TIC for each peak.
Not shown are SIMs for individual peaks, or traces of individual ions over time, which we have also seen.

As a sometime programmer, I am likely to dump the collected data into a matrix (or a spreadsheet), with one dimension reflecting sample times, and the other reflecting the ion in question. For instance, using a spreadsheet, one might have 500 columns reflecting m/z 50 through 550, and however many rows are needed for the sample interval.

One can imagine the possibility of higher achievable sample rates with fewer ions being collected, a full scan offering perhaps less resolution than SIM collection.

Given the complete data, the selection of a particular display is an issue of presentation, not of the data collected.


As rough estimates, assume a 32 bit value (4 bytes) for each sample of 500 ions, giving about 2k bytes of memory per sample. We have about 2600 seconds in a run, and we have isotropic offset of about 150ms. That suggests to me a sampling rate of at least 150/3ms, or 50ms which is 20 samples per second, or 52,000 samples on a 2600 second run. At 2k per sample, that is about 100MB per run.

This kind of data is highly compressible by at least a factor of 10, so I'd expect the complete data for a run to be less than 10Mb. About 70 of these would fit on a CD-ROM. (I'd really expect compression of 100 or so, since most ions will all be zero.)


m said...

What's the evidence that mass spectra, partial mass spectra, selective ion monitoring, etc. were or were not performed by LNND on the Landis samples, were or were not destroyed etc.

Can folks post references to the record here in this thread.

So far I really haven't seen much in the record. I searched the transcript for the term and found very little. Some of the USADA exhibits which I thought showed partial mass spectra that could potentially verify peak identity, I am less sure about now.

Of course, the fact that the mass spectra can't be located in the record doesn't mean they weren't in fact recorded.