Participant:	      Allen
Authors:              Felicity Allen, Tanvir Sajed, Russ Greiner, David Wishart
Affiliations:         Department of Computing Science
		      University of Alberta, Canada

ParticipantID:        FelicityAllenCFMRetrained
Category:             category3
Automatic pipeline:   yes
Spectral libraries:   no

Abstract

We processed the list of molecules and provided candidates using cfm-id.

We combined two scores:  CFM_SCORE + DB_SCORE

CFM SCORE

A new CFM model trained on data from Metlin and NIST MS/MS was used
for the positive mode spectra.  This new model also incorporated
altered chemical features and a neural network within the transition
function.  Mass tolerances of 10ppm were used, and the DotProduct
score was applied for spectral comparisons.  This model only combined
the spectra across energies before training, so only one energy exists
in the output.  For the Negative model we have not yet trained a new
model, so the original negative CFM model was used, as for the CFMOrig
entry.

DB_SCORE

We checked for membership of each candidate in HMDB, ChEBI, a local database of 
plant derived compounds, FOODB and DRUGBANK and assigned +10 to the score for
each database the compound was found to be a member of.

Participant:	      Kind
Authors:              Tobias Kind(1), Hiroshi Tsugawa(2)
Affiliations:         (1) UC Davis Genome Center - Metabolomics, Davis CA, USA 
		      (2) RIKEN Center for Sustainable Resource Science (CSRS), 
		      Wako, Japan

ParticipantID:        tkind
Category:             category3
Automatic methods:    semi-automatic

Abstract
This is a submission for the http://www.casmi-contest.org/2016/
Category 3: Best Automatic Structural Identification - Full Information

This third category uses MS/MS spectra of 208 unknown compounds (validation set).
All MS/MS spectra were obtained on a Q Exactive Plus Orbitrap from Thermo Scientific, 
with <5 ppm mass accuracy and MS/MS resolution of 35,000 using electrospray ionization
and stepped 20/35/50 HCD nominal collision energies. The [M+H]+ (positive) and 
[M-H]- ion masses were recorded. 

A reversed phase C18 column was used (2.6 uM, 2.1x50 mm with 
a 2.1x5 mm precolumn) with a gradient of (A/B): 95/5 at 0 min, 95/5 at 1 min, 
0/100 at 13 min, 0/100 at 24 min (A = water, B = methanol, both with 0.1% formic acid) 
at a flow of 300 uL/min. 

In Category 3 any form of additional information can be used 
(retention time information, mass spectral libraries, patents, reference count.
This allows to demonstrate whether/how much additional information can improve 
the results of the unknown annotation. 

Approach:
Here we used a two-step procedure, first MS-Finder search and then
MS/MS search for confirmation whenever possible.

(1) Molecular formulas and structures were determined with the MS-Finder
software [http://prime.psc.riken.jp/Metabolomics_Software/] by querying
an internal structure databases and all CASMI provided structures. 
Result data was exported as text file and results were formatted for CASMI submission.

First the molecular formulas were determined with Lewis and Senior
check, 97% element ratio check and  20% isotopic abundance ratio 
and 5 ppm mass accuracy for MS1 and 20 ppm for MS2. Elements
CHNOPSFClBrI were included (Si was excluded). The top 5 formula
were regarded for structure queries. However no MS1 information was provided
for this contest, only precursor mass and product ion information.

Each formula was queried  in an internal structure database and the
CASMI compounds for the validation set category 3. An tree-depth 
of 2 and relative abundance cutoff of 1% as well as 
up to 100 possible structures were reported with MS-Finder.

The score was calculated by the in-silico fragmenter that simulates 
the alpha-cleavage of linear chains up to three chemical bonds 
with consideration of the bond-dissociation energy. Multiple bonds 
(double-, triple-, or cycles) are modeled as penalized 
single bonds in which hydrogens are lost. The final score also
includes mass accuracy, isotopic ratio, product ion assignment, 
neutral loss assignment and existence of the compound in the
internal MS-Finder structure databases.

(2) MS/MS search was used for further confirmation and the 
NIST MS Search GUI [http://chemdata.nist.gov/] together with 
major MS/MS databases such as NIST, MONA, ReSpect and MassBank 
was utilized. 

The precursor was set to 5 ppm and product ion search tolerance to 200 ppm.
For some of the searches that gave no MS/MS results also
simple similarity search without precursor info was used.
Results that gave overall low hit scores were also cross-referenced
with the STOFF-IDENT database of environmentally-relevant substances,
to obtain information on potential hit candidates.