/sci/ - Science & Math » Thread #11425017

7KiB, 250x230, 1496611865877.jpg

View Same Google iqdb SauceNAO

Hemolithin: a Meteoritic Protein containing Iron and Lithium

Anonymous Fri 28 Feb 18:18:29 2020 No.11425017 View Reply Original Report

Quoted By: >>11425208 >>11426162

https://arxiv.org/abs/2002.11688v1?fbclid=IwAR3K8-qMf66J3ttplDhEvuTs01pCyUHTViGbE-njfXbI87IxtNR9Jz3ODQQ
>This paper characterizes the first protein to be discovered in a meteorite. Amino acid polymers previously observed in Acfer 086 and Allende meteorites [1,2] have been further characterized in Acfer 086 via high precision MALDI mass spectrometry to reveal a principal unified structure of molecular weight 2320 Daltons that involves chains of glycine and hydroxy-glycine residues terminated by iron atoms, with additional oxygen and lithium atoms.
Thoughts, /sci/?

Brainlet

Brainlet Fri 28 Feb 2020 18:19:31 No.11425020 Report

Quoted By: >>11425138 >>11429697

What does that mean ?
Aliens ?

Anonymous

Anonymous Fri 28 Feb 2020 18:54:28 No.11425138 Report

Quoted By:

>>11425020
probably.

Anonymous

Anonymous Fri 28 Feb 2020 19:16:18 No.11425208 Report

Quoted By: >>11425470

>>11425017
Doesn’t look peer reviewed to me...

Anonymous

Anonymous Fri 28 Feb 2020 20:36:23 No.11425470 Report

Quoted By: >>11426162

>>11425208
Surely they fabricated the mass spectra!

Anonymous

Anonymous Sat 29 Feb 2020 00:34:19 No.11426162 Report

Quoted By: >>11426290 >>11427230

>>11425017
>mass spec
It's bullshit, disregard it.
>>11425470
I don't know what you're even trying to imply. Faking mass spectra is literally just about taking any spectrum of anything you like, randomly selecting as many peaks as you need for your y- and b- ladder, and changing the m/z accordingly. That's assuming you even want to go through that much trouble to make it look believable, otherwise you can just generate the y- and b-series yourself, randomly dropout 20-30% of the y-series, randomly dropout 40-60% of the b-series, and add a few hundred random peaks.
That's not even the problem by the way. The problem is that nothing can be said from the spectrum of anything that's not a histone right now (histone spectra are ridiculously clean but everything else is so full of bullshit it's impossible to get a positive match consistently even if you know what the precursor fragment was, and if you don't know what the precursor was you're even more fucked). For the record, the state of the art methods are hardcore tuned to trypsin-digested proteins, and have huge problems with co-elution. Taking that into account, if you use SPOT to make some clean peptides that are trypsic-like but don't coelute, and pass it through a quadrupole orbitrap (best you can do in 2020), you will miss 80% of proteins in your sample outright. Of the remaining 20%, 50% will be incorrectly identified.
That's in the most optimal setup for mass spectrometry, far removed from realistic samples. If you deviate in any way whatsoever, all error rates go up to like 99.99%.
Not to mention contamination - the fucking flask your sample is in will give you adducts that can appear to match AA m/z's in a SPOT sample, now imagine what happens when you're touching a meteorite.

Anonymous

Anonymous Sat 29 Feb 2020 01:14:36 No.11426290 Report

Quoted By:

>>11426162
based

Anonymous

Anonymous Sat 29 Feb 2020 01:37:13 No.11426348 Report

Quoted By: >>11426350

>centered on 16 residues
SNORE

Anonymous

Anonymous Sat 29 Feb 2020 01:38:20 No.11426350 Report

Quoted By:

>>11426348
Awful. 16 isn't even a small protein, it's a peptide. It can't maintain even a secondary structure.

Anonymous

Anonymous Sat 29 Feb 2020 10:31:01 No.11427230 Report

Quoted By: >>11427642

>>11426162
>most used and best analytical chem instrumentation worldwide is bs

Anonymous

Anonymous Sat 29 Feb 2020 14:19:40 No.11427642 Report

Quoted By: >>11427647

>>11427230
Just because it's used doesn't mean it's any good, only that there are no alternatives.
Moreover it works OK for quant (which requires already knowing your sequence) and for small molecules (MS1 setting). The regime where it breaks is MS2 and MS3, regimes where there's no calibration scheme for the instruments whatsoever (i.e. AA-level peptide sequencing). To do protein, you can't actually just pass the protein in the instrument because it's too big, so you have to digest it e.g. by trypsin. Moreover the instrument can only capture charged fragments, so you usually use ESI (though here they used MALDI) to ionize the fragments.
This already leaves a lot of shit at risk of getting fucked.
Next, as mentioned before, there's no calibration at all. The only time those machines get calibrated is by using a pool of controlled chemicals called a promix, i.e. purely in MS1 mode and purely for small molecules.
After that, the leading computer solutions (PEAKSDB and Andromeda from MaxQuant) are closed source crapware. The result is that PEAKSDB only gives you one (1) match per spectrum (its top guess), meaning you can't even check if another similar peptide was more likely to be good by manual examination. Things like PTMs completely thwart it.
Andromeda does a thing where matches are transferred between runs to "improve" matches, but this only putatively works in generic samples. Mix cultures and you're fucked.
The generic method used to match peptides to protein is just to see where all the peptides match on the proteome and when there are more than one match, take the match with the most peptides attached to it. You can imagine how ridiculously naive that is. In fact it's so naive, that PEAKS added a de novo sequencing algo that doesn't actually work (it literally can't even get one amino acid right except in very clean sequences that don't happen in real life), and that improved their performance significantly by adding spurious peptides to do matches.

Anonymous

Anonymous Sat 29 Feb 2020 14:21:18 No.11427647 Report

Quoted By:

>>11427642
All the metrics I have talked about so far are also readily available either in 3rd party reviews of methods, or in the papers of the relevant softwares (like PEAKS). It's not a secret.

Anonymous

Anonymous Sun 01 Mar 2020 02:42:09 No.11429697 Report

Quoted By:

>>11425020
aliens snaliens

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches