bioinformatics of metaspace, presented at ourcon'16
TRANSCRIPT
![Page 1: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/1.jpg)
METASPACE TRAINING COURSE PART 1. THEORY
Metabolite annotation In HR imaging MS
Theodore alexandrov EMBL / UCSD / SCILS
@thalexandrov
![Page 2: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/2.jpg)
m/z203.103m/z212.852 m/z223.075
C. albicans
E. coli
with Alex Koumoutsi, Nassos Typas @ EMBL
C. albicans
E. coli
metabolite identification
![Page 3: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/3.jpg)
dataset
100 GB 100.000 spectra
10.000.000 images
metabolome
50.000-100.000 molecular structures
Metabolite annotation Of 10.000.000 ion images
in-source fragmentation
ion adducts
m/z A m/z B m/z C
AMP MW 347.063 [M+H]=348.071
For 50K moleculaR
structures Dark matter
isotopologues
![Page 4: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/4.jpg)
Targeted metabolite imaging howto
1. Consider possible adducts – +H, +Na, +K or –H, –Cl
2. Calculate m/z of each adduct – principal or monoisotopic
3. Examine ion images
4. Examine the potential isotopic pattern
5. Estimate the ambiguity (any isomers? isobars?)
6. Validate with in situ MS/MS – On a region of high intensity
![Page 5: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/5.jpg)
our bioinformatics solution
![Page 6: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/6.jpg)
Palmer et al., Nature Methods, accepted
Molecular annotation
Fdr calculation
![Page 7: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/7.jpg)
Molecular annotation
Fdr calculation
1. Consider possible adducts 2. Calculate m/z of each adduct 3. Examine images 4. Examine the potential isotopic pattern 5. Estimate the ambiguity (any isomers? isobars?)
1
3
4
3 2
5
![Page 8: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/8.jpg)
Measure of Spatial chaos
Structured informative
Chaotic non-informative
![Page 9: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/9.jpg)
Spectral & spatial isotope scores
9
monoisotopic image structured? ok fine isotope structure matching theor? ok isotopic images co-localized? not
è doesn‘t pass the filters
![Page 10: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/10.jpg)
Molecular annotation
Fdr calculation
Ok, all ions are scored by their likelihood
... But how to choose the cutoff?
![Page 11: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/11.jpg)
How to choose msm cutoff
![Page 12: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/12.jpg)
how to select parameters in proteomics
Database
Data
MolecularIden=fica=on Listofmolecules
Correct?
1. How to quantify correctness?
2. False Discovery Rate FDR = ratio of false positives
3. Don’t know false positives è cannot calculate FDR
4. Can we estimate it?
true positives
false positives
![Page 13: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/13.jpg)
How to estimate fdr In proteomics
Database
Data
MolecularIden=fica=on
MolecularIDs
Fakedatabase MolecularIDs
“Decoy”
“Target”
FDR # false positives for target
# identifications for target
estimated FDR # target FPs
# target IDs = ≈
# decoy FPs
# target IDs
true positives
false positives false positives
positives:
=
=
TargetsimilartoDecoy
defini3on =
![Page 14: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/14.jpg)
ExplainmelikeI’mfive!
![Page 15: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/15.jpg)
you won tickets to chile!
![Page 16: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/16.jpg)
which cities are sunny?
Whom to
believe?
1. Let’s take cities which we know answers for (always rainy) = decoy
2. Ask to predict what the weather will be like there “sunny” they say è false positive (# decoy FPs)
![Page 17: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/17.jpg)
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR
FDR FDR
HM... How is it related to imaging ms?!
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR FDR FDR FDR
FDR FDR FDR FDR FDR FDR
FDR FDR
FDR FDR FDR
FDR FDR
![Page 18: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/18.jpg)
FDR for imaging ms
Fdr calculation
Palmer et al., Nature Methods, accepted
![Page 19: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/19.jpg)
Fdr-controlled metabolite annotation
Palmer et al., Nature Methods, accepted
![Page 20: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/20.jpg)
Does it really work?
![Page 21: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/21.jpg)
Test case mouse brain sections
Reference H&E images from Allen Brain Atlas
Bregma 1.42mm 3 serial sections
Bregma -1.46mm 1 section
Bregma -3.88mm 1 section
animal 1 animal 2
![Page 22: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/22.jpg)
Imaging ms
DHB matrix ImagePrep 50 μm pixel size
10K spectra 100-1.200 m/z 20-30 GB
► ►
solarix XR 7T Paracell 130K @ m/z 400
with Regis Lavigne, Charles Pineau @ UR1, France
![Page 23: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/23.jpg)
FDR (10%)-controlled annotation OVERVIEW
![Page 24: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/24.jpg)
FDR (10%)-controlled annotation DETECTED METABOLITES
![Page 25: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/25.jpg)
LC-MS/MS validation XIC
![Page 26: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/26.jpg)
LC-MS/MS validation
Standard available
![Page 27: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/27.jpg)
LC-MS/MS validation no standard
![Page 28: Bioinformatics of METASPACE, presented at OurCon'16](https://reader031.vdocuments.mx/reader031/viewer/2022030307/58e7f8dd1a28abf13f8b4ed3/html5/thumbnails/28.jpg)
@ EMBL Andy Palmer, Prasad Phapale Vitaly Kovalev, Sergey Nikolenko Artem Tarasov, Dominik Fay Luca Rappez
Thank you
METASPACE Consortium Christoph Steinbeck, EMBL-EBI
Lennart Martens, VIB Charles Pineau, Regis Lavigne,
URennes Pieter Dorrestein, UCSD Zoltan Takats, Kirill Veselkov, ICL Dennis Trede, SCiLS Oliver Panzer, ERS and their team members
Bruker Michæl Becker, Jens Fuchser
@alexandrovteam
European Horizon2020 HEALTH programme