Biology Reference
In-Depth Information
95%
CCDSgene
85%
REFgene
MGCgene
EXOGENE
75%
UCSC-KNOWNgene
ENSgene
JIGSAW
Single-genome
ab initio
ENSEMBL
PAIRAGON+NSCAN_EST
PAIRAGON-any
Dual- or multiple-genome-based
EST-, mRNA-, & protein-based
All available evidence
65%
55%
ACEVIEW
45%
FGENESH++
AUGUSTUS-EST
NSCAN
35%
AUGUSTUS-any
25%
MARS
ACEMBLY
TWINSCAN
AUGUSTUS-dual
AUGUSTUS-abinit
DOGFISH
15%
SGPgene
ECgene
GENSCAN
GENEZILLA
GENEID
GENEMARK.hmm-B
EXONHUNTER
5%
GENEMARK.hmm-A
SAGA
5%
15%
25%
35%
45%
55%
65%
75%
85%
95%
Sensitivity
Fig. 1.
The human ENCODE Genome Annotation Assessment Project (EGASP)
10
benchmarking analysis of the gene prediction accuracy (sensitivity versus specificity at
the gene level) of some of the most popular gene prediction software packages.
Knowledge-based approaches, which incorporate multiple lines of evidence, tend to
perform better than
ab initio
methods, which can nevertheless provide complemen-
tary information for more complete genome annotations.
last several years despite the application of extensive resources. As it is not
feasible to allocate comparable resources to characterize other incoming
eukaryotic genomes, including some of key medical importance, the
major hope remains to improve computational approaches and combine
them with high-throughput experimental screens.
11
3. Protein Families
Homology is a general concept in biology that recognizes similarities
between entities as evidence of shared ancestry. This premise is particularly
