Charges in raw arrays

[RienLeuvenink]

Hi,

I am trying to look at some ion data but I can't seem find the charges. I've included an example spectrum below which has annotated charges when I open the raw file in freestyle. I assume this is encoded in the raw file and not something that freestyle makes up..
below how I access the file. Maybe I'm just overlooking something

`
let mut ms_file =
MZReader::open_path(r"D:\MS data\2025-01 January\ETD test\MS1_example.raw").unwrap();

dbg!(ms_file
    .get_spectrum_by_index(0)
    .unwrap()
    .raw_arrays()
    .unwrap());

`

https://filesender.surf.nl/?s=download&token=168ebd63-923e-49f3-8424-b65c3a5e942c

[mobiusklein]

Actually, Freestyle, like Excalibur before it, does include a form of isotopic pattern fitting and charge state prediction that it does on the fly, and that charge information may not make its way to other files. I'll still check to see if I'm missing something in the Thermo RAW file parsing process for your file though. Charge states aren't (or weren't) stored in the raw file because how you compute them depends heavily on what you're analyzing. Thermo used to be terrible at monoisotopic peak prediction for larger molecules when assigning precursors, but in 2018 or so introduced an improved model for peptide-like ions, called "peptide match" in the instrument method. Such a model wouldn't work properly on an ion from a wildly different class of molecules like say a highly sulfated heparan because of how the "average monomer" scaling process that general-purpose isotopic pattern predictors use differ between the two classes. This is why I wrote `ms_deisotope` and later `mzdeisotope` (I just did a big bug fix release of this to fix issues with reading deconvolved peaks back out of mzML files), and why other informatics people in the academic lab I did my PhD in wrote molecule-class specific deconvolution software Thank you, Joshua Klein

…

On Fri, Jan 10, 2025 at 8:55 AM RienLeuvenink ***@***.***> wrote: Hi, I am trying to look at some ion data but I can't seem find the charges. I've included an example spectrum below which has annotated charges when I open the raw file in freestyle. I assume this is encoded in the raw file and not something that freestyle makes up.. below how I access the file. Maybe I'm just overlooking something ` let mut ms_file = MZReader::open_path(r"D:\MS data\2025-01 January\ETD test\MS1_example.raw").unwrap(); dbg!(ms_file .get_spectrum_by_index(0) .unwrap() .raw_arrays() .unwrap()); ` https://filesender.surf.nl/?s=download&token=168ebd63-923e-49f3-8424-b65c3a5e942c — Reply to this email directly, view it on GitHub <#20>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAK4E6I67BHRKHMA6NOC6ST2J7GNJAVCNFSM6AAAAABU6NXZ5WVHI2DSMVQWIX3LMV43ASLTON2WKOZSG44DAMRUG4YTEMQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[RienLeuvenink]

Thanks for your quick response. A collaborator of mine uses the pymsfilereader for python and I've been told that he can access the charge states through this. I see that you where a collaborator on this too, I also see that this package uses the MSFileReader instead of the RawFileReader maybe this is where the difference stems from?

Kind regards,
Rien

[mobiusklein]

Yes, the old Thermo MSFileReader C++/COM library included both a peak picking and a charge deconvolution algorithm, but neither are easily configurable.

RawFileReader does seem to include a few charge state estimators. One is based upon the THRASH algorithm from Horn et. al. and seems to work with the peak data stream directly, while others appear to be tied to some other GUI-facing application and don't have a clear way of interacting with the CentroidStream API.

Additionally, your file and some of my newer test files seem to have charge annotations, but they don't include isotopic pattern assignments so it would effectively say "this peak is predicted to be this charge state, but whether it is a monoisotopic peak is unknown". This is representable with mzpeaks::DeconvolutedPeak, but it's not what it's meant for, semantically. If this is still desirable, I can add the plumbing to thermorawfilereader to surface this information.

[RienLeuvenink]

If possible from your side I'd be keen to try it out and compare it to what freestyle tells me. In my experience and for the type of proteins we're looking at freestyle does quite a decent job of assigning charge-states etc.

[mobiusklein]

I've added support for extracting the charge array to the main branch:

use std::io;

use mzdata::prelude::*;

fn main() -> io::Result() {
    let mut reader = ThermoRawReader::open_path("./test/data/small.RAW")?;
    reader.set_centroiding(true);
    reader.set_load_extended_spectrum_data(true);
    
    let spec = reader.get_spectrum_by_index(0).unwrap();
    let arrays =  spec.arrays.as_ref().unwrap();
    let z_array = arrays.charges().unwrap();
    
    Ok(())
}

Again, this charge array doesn't deisotope the spectrum. I think it just does a variation of the Patterson charge state determination method like what's shown in the Horn et al. paper to determine the charge state from peak spacing, and may be incorrect when you have interleaved isotopic peaks.

The change also includes other data arrays for FT spectra including a signal-to-noise-ratio array and a raw baseline signal array. I am just pulling what's available from the GetAdvancedPacketData function at https://github.com/mobiusklein/thermorawfilereader.rs/blob/main/librawfilereader/Lib.cs#L736-L840.