Table 1.

Annual growth of mammalian and human RefSeq transcript records

Releasea	Taxab	Mammalian records				Human records
		Total transcripts	Percent modelsc	Percent curatedd	Percent ncRNAe	Total transcripts	Total genesf	Percent modelsc	Percent curatedd	Percent ncRNAe
1	5	126 980	68	7	<1	38 556	na	50	22	<1
6	10	79 686	41	17	<1	28 176	na	23	42	<1
12	19	158 111	65	11	<1	29 490	na	18	50	1
18	28	263 628	77	7	5	40 342	28 514	38	39	2
24	35	338 204	80	7	8	38 709	29 398	34	46	14
30	37	340 968	77	9	9	45 511	27 741	41	45	16
36	42	346 976	74	12	8	43 589	29 071	30	60	13
42	42	425 170	76	12	9	46 111	29 954	27	63	15
48	43	470 979	76	12	9	46 912	27 619	20	70	25
54	45	515 900	76	12	9	44 951	26 440	10	79	23
60	59	1 263 067	90	5	6	47 619	26 266	11	79	24

^aRelease numbers listed correspond to ∼12 month intervals beginning from the first release in June 2003. The number of human transcripts in release 60 (July 2013) reflects the November 2012 genome annotation of three assemblies (GRCh37.p10, HuRef and CHM1_1.0) plus records added through ongoing curation activities.

^bThe number of distinct NCBI Taxonomy IDs included in the RefSeq vertebrate_mammalian FTP directory that have a publicly available nuclear genome records. Twelve taxa are represented by un-annotated ENCODE genomic region records only. Mammals for which only a mitochondrial genome sequence is available are excluded. Data reported in Table 1 were extracted from archived reports available at ftp://ftp.ncbi.nlm.nih.gov/refseq/release/release-catalog/ using files named as ‘RefSeq-release##.catalog.gz’.

^cThe percent of total transcripts that are model RefSeqs (with XM or XR accession prefix) generated by NCBI’s eukaryotic annotation pipeline. The percent known RefSeqs (with NM or NR prefix) can be inferred from this value (100% − percent model RefSeqs = percent known RefSeqs).

^dThe percent of total transcripts that are known RefSeq records that have been curated by NCBI staff and are annotated with a ‘validated’ or ‘reviewed’ status in the COMMENT block of the RefSeq record. Validated records have undergone sequence review by NCBI staff, whereas a reviewed record includes curation of descriptive information, such as names, publications and a RefSeq summary in addition to sequence review. Known RefSeq records that have not been curated are not included; thus, the number of model records and curated records do not sum to 100%.

^eThe percent of total transcripts that are not protein coding. This includes model or known long non-coding RNAs (lncRNA), small RNAs (e.g. microRNA, snoRNA, etc.), ribosomal RNAs and transcribed pseudogenes. Transfer RNAs, which are annotated on genomic records using tRNAscan but not tracked with RefSeq accessions, are not included.

^fThe number of human genes per release was derived using FTP files named as ‘release##.accession2geneid.gz’. This file was not provided prior to release 14.