EVALUATION PROTOCOL

Overview of Sensitivity and Specificity Scores

The accuracy measures being used along this section are described in Burset and Guigó (Genomics, 34/3:353-357, 1996), Reese et al (Genome Research, 10/4:483-501, 2000) and Guigó et al (Genome Research, 10/10:1631-1642, 2000). Figure above was adapted from the first reference (Burset and Guigó, 1996). Evaluation at gene and transcript level for the results displayed on this site was performed following the EGASP “IMIM” criteria, which is described on the supplementary web materials for Guigó et al (Genome Biology, 7/Suppl1:S2, 2006).

Computational Protocol and Software

Go to the analysis protocol and software page

RGASP: ROUND 1

Codes for Participants

Evaluation Scores Data

Hub of Graphical Results

NOTES:

• The submissions were split into four categories: partial, CDS-only, exon-only and full-predictions. The pages linked from the table summarize, for every evaluation variable taken into consideration, a plot with all submissions together (except for the last row boxplots), followed by four plots, one per each of the mentioned categories. When data is not available for a given variable, an asterisk will mark the corresponding submission, variable or sequence boxplot. When there was no data available for all the elements within a given category on a given variable, an empty plot is shown as placeholder. In example, exon-only predictions do not contain CDS features, thus you get an empty plot for CDS related variables on that category.
• All scatterplots show data on every submission and every sequence. Each dot corresponds to the SN vs SP on a given sequence for a given submission, no labels though. To check those details, you can look at the box-plots.
• All boxplots from the above table show (SN+SP)/2 on the y-axis. Box upper and bottom edges correspond to the 3rd and 1st quartiles, the line in the middle is the median. The ticks above and below define the range of two times the interquartile range, which means that points out of those marks can be considered as outlayers. The points marked as outlayers only show their label when they refer to sequences, as the names for the submission files or evaluation variables are too long to display on those plots

RGASP: ROUND 2

Codes for Participants

Annotation Datasets Feature Summaries

Evaluation Scores Data

Hub of Graphical Results

NOTES: Same points from the previous NOTES on the section about RGASP round 1 apply to the results linked from this table.