NCBI Homo sapiens Annotation Release 107

The RefSeq genome records for Homo sapiens were annotated by the NCBI Eukaryotic Genome Annotation Pipeline, an automated pipeline that annotates genes, transcripts and proteins on draft and finished genome assemblies. This report presents statistics on the annotation products, the input data used in the pipeline and intermediate alignment results.

The annotation products are available in the sequence databases and on the FTP site.

This report provides:

• Annotation Release information: The name of the release, important dates, the software version
• Assemblies: A brief description of the annotated assembly(ies)
• Gene and feature statistics: The counts and characteristics of the annotated features
• Alignment of the annotated proteins to a set of high-quality proteins: The number of annotated proteins with hits to a set of high-quality proteins
• Masking of genomic sequence: How much of the genome was masked
• Transcript and protein alignments: The number and type of evidence retrieved from public databases and used for gene prediction
• Similarity of current and previous assembly: The similarity of the current and previous assembly
• Similarity of alternate and reference assemblies: The similarity of the alternate assemblies to the reference

For more information on the annotation process, please visit the NCBI Eukaryotic Genome Annotation Pipeline page.

Annotation Release information

This annotation should be referred to as NCBI Homo sapiens Annotation Release 107

Annotation release ID: 107
Date of Entrez queries for transcripts and proteins: Feb 19 2015
Date of submission of annotation to the public databases: Mar 12 2015
Software version: 6.2

Assemblies

The following assemblies were included in this annotation run:

Assembly name	Assembly accession	Submitter	Assembly date	Reference/Alternate	Assembly content
GRCh38.p2	GCF_000001405.28	Genome Reference Consortium	12-05-2014	Reference	25 assembled chromosomes; unplaced scaffolds
CHM1_1.1	GCF_000306695.2	Washington University School of Medicine	06-14-2013	Alternate	23 assembled chromosomes

Gene and feature statistics

Counts and length of annotated features are provided below for each assembly.

Feature counts

Feature	GRCh38.p2	GRCh38.p2 Primary Assembly	GRCh38.p2 All Alt Loci	GRCh38.p2 PATCHES	CHM1_1.1
Genes and pseudogenes	51,087	50,856	2,322	285	49,687
protein-coding	20,576	20,486	1,029	116	20,266
non-coding	18,037	17,940	661	80	17,675
pseudogenes	12,474	12,430	632	89	11,746
genes with variants	20,098	20,004	721	106	14,369
placed on multiple assembly-units	2,276	na	649	na	na
mRNAs	98,108	94,414	4,875	460	39,983
fully-supported	97,809	94,140	4,854	456	39,747
with > 5% ab initio	217	199	15	3	178
partial	42	41	201	13	643
with filled gap(s)	0	0	0	0	0
placed on multiple assembly-units	1,607	na	580	na	na
known RefSeq (NM_)	38,668	38,611	1,512	186	38,372
model RefSeq (XM_)	59,440	55,803	3,363	274	1,611
Other RNAs	41,516	39,401	2,205	204	33,037
fully-supported	39,773	38,057	1,885	203	31,699
with > 5% ab initio	0	0	0	0	0
partial	7	7	36	2	96
with filled gap(s)	0	0	0	0	0
placed on multiple assembly-units	288	na	104	na	na
known RefSeq (NR_)	9,861	9,857	335	41	9,651
model RefSeq (XR_)	29,925	28,213	1,550	162	22,065
CDSs	98,652	94,800	5,031	457	40,357
fully-supported	97,809	94,140	4,854	456	39,747
with > 5% ab initio	253	232	18	3	203
partial	42	41	172	9	333
with major correction(s)	108	65	54	6	496
known RefSeq (NP_)	38,668	38,611	1,510	183	38,370
model RefSeq (XP_)	59,440	55,803	3,363	274	1,611

Detailed reports

Alignment of the annotated proteins to a set of high-quality proteins

The final set of annotated proteins was searched with BLASTP against the UniProtKB/Swiss-Prot curated proteins, using the annotated proteins as the query and the high-quality proteins as the target. Out of 20102 coding genes in the GRCh38.p2 primary assembly, 19542 genes had a protein with an alignment covering 50% or more of the query and 18934 had an alignment covering 95% or more of the query. Out of 19889 coding genes in the CHM1_1.1 assembly, 19381 genes had a protein with an alignment covering 50% or more of the query and 18761 had an alignment covering 95% or more of the query.

Definition of query and target coverage. The query coverage is the percentage of the annotated protein length that is included in the alignment. The target coverage is the percentage of the target length that is included in the alignment.

Below is a cumulative graph displaying the number of genes with alignments above a given query or target coverage threshold.

Query: annotated proteins
Target: UniProtKB/Swiss-Prot curated proteins

Masking of genomic sequence

Transcript and protein alignments are performed on the repeat-masked genome. Below are the percentages of genomic sequence masked by WindowMasker and RepeatMasker for each assembly. RepeatMasker results are only used for organisms for which a comprehensive repeat library is available.

For this annotation run, transcripts and proteins were aligned to the genome masked with RepeatMasker only.

Assembly name	Assembly accession	% Masked with RepeatMasker	% Masked with WindowMasker
GRCh38.p2	GCF_000001405.28	49.29%	36.10%
CHM1_1.1	GCF_000306695.2	47.21%	33.97%

Transcript and protein alignments

The annotation pipeline relies heavily on alignments of experimental evidence for gene prediction. Below are the sets of transcripts and proteins that were retrieved from Entrez, aligned to the genome by Splign or ProSplign and passed to Gnomon, NCBI's gene prediction software.

Depending on the other evidence available, long 454 reads (with average length above 250 nt) may be aligned as traditional evidence and reported in the Transcript alignments section or aligned with short reads and reported in the Short read transcript alignments section.

Transcript alignments

RefSeq transcript alignment quality report

The known RefSeq transcripts (NM_ and NR_ accessions) are a set of hiqh-quality transcripts maintained by the RefSeq group at NCBI. Alignment statistics for this group of transcripts, such as percent and number of sequences not aligning at all, percent best alignments split between multiple scaffolds, and percent alignments not covering the full CDS are indicative of the genome quality and are provided below.

	GRCh38.p2 Primary Assembly	CHM1_1.1 Primary Assembly
Number of sequences retrieved from Entrez	49,683	49,683
Number (%) of sequences not aligning	18 (0.04%)	84 (0.17%)
Number (%) of sequences with multiple best alignments (split genes)	8 (0.02%)	36 (0.07%)
Number (%) of sequences with CDS coverage < 95%	15 (0.04%)	265 (0.69%)

Short read transcript alignments

The following short reads (RNA-Seq) from the Sequence Read Archive were also used for gene prediction:

  Hide alignments statistics, by sample (SAME, SAMN, SAMD, DRS)  Show alignments statistics, by run (ERR, SRR, DRR)

Protein alignments

Assembly-assembly alignments of current to previous assembly

When the assembly changes between two rounds of annotation, genes in the current and the previous annotation are mapped to each other using the genomic alignments of the current assembly to the previous assembly so that gene identifiers can be preserved. The success of the remapping depends largely on how well the two assembly versions align to each other.

Below are the percent coverage of one assembly by the other and the average percent identity of the alignments. The 'First pass' alignments are reciprocal best hits, while the 'Total' alignments also include 'Second pass' or non-reciprocal best alignments. For more information about the assembly-assembly alignment process, please visit the NCBI Genome Remapping Service page.

First Pass	Total
GRCh38.p2 (Current) Coverage: 99.31%	GRCh38.p2 (Current) Coverage: 99.99%
GRCh38 (Previous) Coverage: 99.71%	GRCh38 (Previous) Coverage: 99.99%
Percent Identity: 99.99%	Percent Identity: 99.99%

Assembly-assembly alignments of alternate to reference assembly

When multiple assemblies of good quality are available for the organism, the annotation of all is done in coordination. The alternate assemblies are aligned to the reference assembly and the best reciprocal best hits are used to identify corresponding regions, that can then be annotated together.

Below are the percent coverage of one assembly by the other and the average percent identity of the alignments. The 'First pass' alignments are reciprocal best hits, while the 'Total' alignments also include 'Second pass' or non-reciprocal best alignments. For more information about the assembly-assembly alignment process, please visit the NCBI Genome Remapping Service page.

First Pass	Total
CHM1_1.1 (Alternate) Coverage: 99.64%	CHM1_1.1 (Alternate) Coverage: 99.83%
GRCh38.p2 (Reference) Coverage: 92.05%	GRCh38.p2 (Reference) Coverage: 96.64%
Percent Identity: 99.79%	Percent Identity: 99.75%

References

• RefSeq: Pruitt KD, Brown GR, Hiatt SM, Thibaud-Nissen F, Astashyn A, Ermolaeva O, Farrell CM, Hart J, Landrum MJ, McGarvey KM, Murphy MR, O'Leary NA, Pujar S, Rajput B, Rangwala SH, Riddick LD, Shkeda A, Sun H, Tamez P, Tully RE, Wallin C, Webb D, Weber J, Wu W, Dicuccio M, Kitts P, Maglott DR, Murphy TD, Ostell JM. Nucleic Acids Research 2014, 42(Database issue):D756-63
• RepeatMasker: Smit AFA, Hubley R, Green P. RepeatMasker Open-3.0. 1996–2004. http://www.repeatmasker.org
• WindowMasker: Morgulis A, Gertz EM, Schäffer AA, Agarwala R. Bioinformatics 2006, 2:134-41
• Splign: Kapustin Y, Souvorov A, Tatusova T, Lipman D. Biology Direct 2008, 3:20