Isorix calibration inquiry

[Libesha Anparasan]

Hi Alex!

Hope you are doing well!

My name is Libesha and I am currently using Isorix to compare the hydrogen isotopic fidelity of different fatty acids to precipitation in north America in butterflies. The calibration function provided with the standard work flow uses a SpaMM summary output which means (to my understanding) that the only metric of really measuring or comparing the models to each other is the phi value. I am reaching out as I am wondering if there is a way to extract the values that are used to make the calibration plot (in particular the precipitation ones) so that I can just plot the values myself to assess via linear regression etc?  

Thank you in advance for your time!

Best,

Libesha Anparasan

[Alexandre Courtiol]

Hi Libesha,

sorry for the delay and a warm welcome to this group.

Which models are you trying to compare? Calibration models?

If so, IsoriX does not offer flexibility in those and try to do the best it can given the setup (provided by the argument "method" to `calibfit()`).

In general, comparing models using their residual variance only may not be a good idea.

Assuming that [calibfit_object] is the name of the object created by your call to `calibfit()`, you can easily fetch the fitted model (`[calibfit_object]$calib_fit`).

So, in principle, you could thus run any diagnostics you would run on a LMM (e.g. `AIC()`).

(The fitting procedure is a little complicated in the background (see `objective_fn_calib()` in the source code), so if something do not look right when applied to the fitted object, that could be it (I would need to dig to know for sure).)

Under the default method, the precipitation value used in the calibration plot are predicted by the isoscape.

Those can be access using `[calibfit_object]$data$mean_source_value`, so you can use those for plotting things yourself.

I hope this helps.

++

[Libesha Anparasan]

Hi again!

Yes I am comparing the calibration models of different fatty acids with precip. AIC did work and I was able to access the precip values for plotting myself as well! Thank you very much