Calibrating E-values for MS2 database search methods

Biology Direct

Table 2 Scaling of the database size and functional transformation to yield calibrated E-values.

Method	reported	x _db	x_1gr = x_db (1gr/db_size)^α	E = f(x_1gr)
SEQUEST	X-correlation	x_db = e^{-(X-correlation)}	x_1gr = x_db (1gr/db_size)^-0.176	f(x_1gr) = e^10.59 x_1gr^4.11
InsPecT	MQ_score	x_db = e^-(MQ_score)	x_1gr = x_db (1gr/db_size)^-1	f(x_1gr) = [e^16.08 x_1gr ^1.4]^1/2 if x_1gr ≥ 0.148
				f(x_1gr) = [e^2.78 x_1gr ^0.48]^1/2 if x_1gr > 0.148
ProbID	Posterior Prob.	x_db = e^{-(Posterior Prob.)}	x_1gr = x_db (1gr/db_size)^-5	f(x_1gr) = e^13.82 x_1gr ^0.17
Mascot	E-value	x_db = E-value	x_1gr = x_db (1gr/db_size)^0.301	f(x_1gr) = x_1gr/225
X!Tandem	E-value	x_db = E-value	x_1gr = x_db	f(x_1gr) = x_1gr/10
OMSSA	E-value	x_db = E-value	x_1gr = x_db (1gr/db_size)^-0.301	f(x_1gr) = e^-2.26 x_1gr ^0.57
RAId_Dbs	E-value	x_db = E-value	x_1gr = x_db	E = f(x_1gr) = x_1gr

Scaling of the database size and functional transformation to yield calibrated E-values. The calibration is done in two steps: first transform a database-size-dependent variable x_db (e.g. e^{-(X-correlation)} for SEQUEST and E-value for X!Tandem) into a standard one, x_1gr, which is calibrated for database size of 1 giga residues; second, a functional transformation f(x_1gr) to yield the correct E-value given the equivalent variable x_1gr. The symbol db_size, as suggested by its name, represents the size of the database used in a search. Note that the size scaling and the functional transformations may vary from laboratory to laboratory. However, once calibrated, it becomes sensible to compare the search results from different laboratories using different machines and different analysis software.

ISSN: 1745-6150