ReSpect for Phytochemicals

Polarity & Adduct Query Fragment m/z tolerance (0–10.00) ±	Search Options Showing MS/MS fragmentation association rule. Fragment Filters Relative intensity (0–100.0) ≤% Ranking (1–99) Top
Reset

Search Example

The following separators are available:

\s	(Whitespace character)
\n	(Carriage return character, Carriage return code)
\t	(Tab)
;	(Semicolon)
:	(Colon)

Example 1 (Tryptophan)

146.0634 159.0952 188.0755 205.0977
Example 2 (Tryptophan)

146.0634;159.0952;188.0755;205.0977;

Fragment Search

Fragment search is a spectrum search, which is based on only the matching of m/z values in MS/MS data.

Query data : m/z values of your MS/MS data

Record : m/z value of ReSpect record

1. Execution of spectrum search using query and record

MS/MS data of query and record	: Q, R
m/z value of query and record	: Q_x, R_x
Match	: \|Q_x1 - R_x1\| ≤ m/z tolerance
Match	: \|Q_x2 - R_x2\| ≤ m/z tolerance
Mismatch	: \|Q_x3 - R_x3\| > m/z tolerance

Fragment search using query and record Graph

MS/MS fragmentation association rule (September 2011)

1. Setting for ReSpect record selection

Relative intensity : ≤20%

Ranking : Top 10

Total : 8,623 ReSpect records

2. Calculation of MS/MS fragmentation association rule

2-1. Using association rule algorithm (see →), total 4,063,748 rules were enumerated.

2-2. After the calculation of association rules, the 12,220 of MS/MS fragmentation association rules are defined as followed.

In a rule, the antecedent is a single and max value of m/z in the rule.

Difference of m/z values between antecedent and consequent is more than 10.

Transactions are grouped by same polarity and adduct type in a compound class level 1.

Table

**Table.** Transaction of MS/MS fragmentation association rules in ReSpect records.
Compound class level 1	Polarity	Adduct type	Transaction
Alkaloid	Negative	[M]^-	6
Alkaloid	Negative	[M-H]^-	37
Alkaloid	Positive	[M+H]⁺	354
Amino acid	Negative	[M-H]^-	160
Amino acid	Positive	[M+H]⁺	700
Flavonoid	Negative	[M-2H]^-	32
Flavonoid	Negative	[M-Cl]^-	141
Flavonoid	Negative	[M-H]^-	1,369
Flavonoid	Positive	[M]⁺	35
Flavonoid	Positive	[M+H]⁺	1,332
Flavonoid	Positive	[M+Na]⁺	12
Glucosinolate	Negative	[M-H]^-	170
Glucosinolate	Positive	[M+H]⁺	48
Other	Negative	[M-H]^-	1,043
Other	Positive	[M]⁺	5
Other	Positive	[M+H]⁺	1,735
Other	Positive	[M+H]⁺or[M+NH4]⁺	4
Other	Positive	[MH+]⁺	4
Other	Positive	[M+Na]⁺	26
Other	Positive	[M+NH4]⁺	9
Phenylpropanoid	Negative	[M-H]^-	319
Phenylpropanoid	Positive	[M+H]⁺	187
Stilbenoid	Negative	[M-H]^-	24
Stilbenoid	Positive	[M+H]⁺	18
Tannin	Negative	[M-H]^-	102
Tannin	Positive	[M+H]⁺	31
Terpenoid	Negative	[M-H]^-	318
Terpenoid	Positive	[M+H]⁺	293
Terpenoid	Positive	[M+H]⁺or[M+NH4]⁺	45
Terpenoid	Positive	[M+Na]⁺	52

2-3. Calculation of index values for a rule.

Number of transaction including antecedent (X)

Number of transaction including consequent (Y)

Number of all transaction (Z)

Support (X) = count (X) ÷ count (Z)

Support (Y) = count (Y) ÷ count (Z)

Support (X⇒Y) = count (X∪Y) ÷ count (Z)

Confidence (X⇒Y) = count (X∪Y) ÷ count (X)

Lift (X⇒Y) = Confidence (X⇒Y) ÷ Support (Y)

Example evaluation of a rule (A⇒B)
Support (A⇒B)	Frequency of A⇒B can estimate as percentage, e.g., 0–100%.
Confidence (A⇒B)	Strength of A⇒B can estimate as percentage, e.g., 0–100%.
Lift (A⇒B)	Independence of A⇒B can estimate as criterion variable, e.g., 0–∞.
Lift values greater than 1	Frequency of a rule has positively effect on frequency of antecedent.
Lift value is less than 1	Frequency of a rule has negatively effect on frequency of antecedent.
Lift value close to 1	Frequency of a rule has no effect on frequency of antecedent.

▲ Page Top

Association rule algorithm

The association rule is an algorithm of frequent pattern mining for discovering interesting relations between variables in database (Agrawal et al., 1993).

An association rule "milk ⇒ bread, egg" in selected dataset can be predicted as follows: if a customer brought a milk, then also bought bread and egg.

Thus, the left and right hand side in a rule are named antecedent (if) and consequent (then), and these dataset are named transaction.

The index values of the association rules are calculated based on the number of transaction including a specific combination of antecedent and consequent, and the values (Support, Confidence and Lift) are used for evaluation of frequency, strength, and independence.

Reference
Agrawal, R., Imielinski, T., Swami, A., 1993. Mining Association Rules Between Sets of Items in Large Databases. SIGMOD Conference, 207–216.

Example analysis of association rule

Transaction id	Item purchased
1	Milk, Bread, Egg, Rice
2	Juice, Potato chips
3	Bread, Butter
4	Milk, Bread, Egg
5	Milk, Bread
6	Milk, Bread, Egg

Milk⇒Bread, Egg

Support(Milk⇒Bread, Egg)

= count(Milk∪Bread, Egg) ÷ count(Transaction)

= 3÷5

= 0.6

Confidence(Milk⇒Bread, Egg)

= count(Milk∪Bread, Egg) ÷ count(Milk)

= 3÷4

= 0.75

Lift(Milk ⇒Bread, Egg)

= Confidence(Milk⇒Bread, Egg) ÷ Support(Bread, Egg)

= Confidence(Milk⇒Bread, Egg) ÷ count(Bread, Egg) ÷ count(Transaction)

= 0.75÷0.5

= 1.5

Juice⇒Potato chips

Support(Juice⇒Potato chips)

= count(Juice∪Potato chips) ÷ count(Transaction)

= 1÷6

= 0.17

Confidence(Juice⇒Potato chips)

= count(Juice∪Potato chips) ÷ count(Juice)

= 1÷1

= 1

Lift(Juice⇒Potato chips)

= Confidence(Juice⇒Potato chips) ÷ Support(Potato chips)

= Confidence(Juice⇒Potato chips) ÷ count(Potato chips) ÷ count(Transaction)

= 1÷0.17

= 5.88

ReSpect	Fragment Search	Spectrum Search	Keyword Search	Record Index	ACCESSION
ReSpect	Download	Example Analyses	Literatures	Development	Help is displayed.

Query data	: m/z values of your MS/MS data
Record	: m/z value of ReSpect record

Relative intensity	: ≤20%
Ranking	: Top 10
Total	: 8,623 ReSpect records

Search Example

Fragment Search

1. Execution of spectrum search using query and record

MS/MS fragmentation association rule (September 2011)

1. Setting for ReSpect record selection

2. Calculation of MS/MS fragmentation association rule