MScanner: a classifier for retrieving Medline citations

BMC Bioinformatics

Table 1 Feature scores for PG07.

Score	R	$\bar{R}$	p(F_i= 1\|R)	p(F_i= 1\| $\bar{R}$ )	z _i	Type	Term String
8.27	56	140	3.5E-2	9.0E-6	1.3E-5	issn	1744–6872 (Pharmacogenet. Genomics)
7.36	137	855	8.6E-2	5.5E-5	6.4E-5	issn	0960-314X (Pharmacogenetics)
7.24	41	287	2.6E-2	1.8E-5	2.1E-5	issn	1470-269X (Pharmacogenomics J.)
6.85	6	62	3.8E-3	4.0E-6	4.4E-6	mesh	Organic Anion Transport Polypeptide C
5.95	20	509	1.3E-2	3.3E-5	3.4E-5	issn	1462–2416 (Pharmacogenomics)
5.88	31	847	1.9E-2	5.4E-5	5.6E-5	mesh	Steroid 16-alpha-Hydroxylase
5.84	70	1986	4.4E-2	1.3E-4	1.3E-4	mesh	Cytochrome P-450 CYP2D6
5.84	2	57	1.3E-3	3.7E-6	3.8E-6	mesh	Glucuronic Acids
5.79	13	390	8.2E-3	2.5E-5	2.6E-5	mesh	Mephenytoin
5.78	114	3434	7.1E-2	2.2E-4	2.3E-4	mesh	Pharmacogenetics
5.69	1	33	6.3E-4	2.1E-6	2.2E-6	mesh	Methenyltetrahydrofolate Cyclohydrolase
5.54	7	268	4.4E-3	1.7E-5	1.8E-5	mesh	Xeroderma Pigmentosum Group D Protein
5.53	2	78	1.3E-3	5.0E-6	5.1E-6	mesh	Methylthioinosine
5.42	5	216	3.1E-3	1.4E-5	1.4E-5	mesh	Organic Anion Transporters, Sodium-Independent

The support scores for feature occurrence, for retrieval using PG07. "R" denotes |R ∩ Fi = 1| and " $\bar{R}$ " denotes | $\bar{R}$ ∩ Fi = 1|, which are the number of example and rest-of-Medline articles with each feature. p(Fi = 1|R) and p(Fi = 1| $\bar{R}$ are the posterior probabilities for feature occurrence, using zi as the prior count.

ISSN: 1471-2105