This track provides a visual representation of data quality scores, which range from 0 to 1, for each plate in the UW/Regulome experiments. It is one of a set of tracks that annotate continuous DNaseI sensitivity measurements and DNaseI hypersensitive sites (HSs) over ENCODE regions. DNaseI has long been used to map general chromatin accessibility and the DNaseI "hyperaccessibility" or "hypersensitivity" that is a universal feature of active cis-regulatory sequences. The data were produced using quantitative chromatin profiling (QCP) (Dorschner et al., 2004).
Quality scores are available on the following cell lines/phenotypes and chromosomes:
Cell Line/Phenotype Chromosomes CACO2 5, 7, 9, 11, 12, 16, X GM06990 2, 5, 7, 8, 9, 11, 12, 16, 18, X SKNSH 5, 7, 9, 11, 12, 16, X Huh7 2, 8, 11, 18 HepG2 11 K562 11 Adult Erythroblast 11
See the UW/Reg Amplicon track for more information on the cell lines/phenotypes studied in these experiments.
Plates with scores greater than or equal to 0.5 were conservatively considered acceptable for reliable scoring of HSs. Scores are shown in greyscale, with darker colors indicating higher scores.
This composite annotation track consists of several subtracks that show the quality scores for each cell line/phenotype. To show only selected subtracks, uncheck the boxes next to the tracks you wish to hide. The display may also be filtered to show only those items with unnormalized scores that meet or exceed a certain threshhold. To set a threshhold, type the minimum score into the text box at the top of the description page.
Color differences among the subtracks are arbitrary; they provide a visual cue for distinguishing the different cell lines/phenotypes.
QCP was performed as described in Dorschner et al. See the UW/Reg Amplicon track description for more information. QCP assays were formatted into 384-well plates for high-throughput real-time PCR. Each plate was treated as a separate experiment.
Plate quality scores were computed using a Support Vector Machine (SVM). Trained operators manually scored 500 plates, classifying each on a scale of 1 to 5 to rank the degree of experimental noise. The unified set was then used to train an SVM to classify and score "good" and "bad" plates. Good plates were conservatively assigned noise scores of 1 - 3; bad plates received scores of 4 - 5. By performing cross validation on a 90% subsample of the training set, the SVM achieved an ROC (receiver operating characteristic) score of 0.93.
See the UW/Reg Amplicon track description for verification information.
Data generation, analysis, and validation were performed jointly by groups at Regulome Corporation and the University of Washington (UW) in Seattle.
Regulome Corp.: Michael O. Dorschner, Richard Humbert, Peter J. Sabo, Anthony Shafer, Jeff Goldy, Molly Weaver, Kristin Lee, Fidencio Neri, Brendan Henry, Mike Hawrylycz, Paul Tittel, Jim Wallace, Josh Mack, Janelle Kawamoto, John A. Stamatoyannopoulos.
UW Medical Genetics: Patrick Navas, Man Yu, Hua Cao, Brent Johnson, Ericka Johnson, George Stamatoyannopoulos.
UW Genome Sciences: Scott Kuehn, Robert Thurman, William S. Noble.
Dorschner, M.O., Hawrylycz, M., Humbert, R., Wallace, J.C., Shafer, A., Kawamoto, J., Mack, J., Hall, R., Goldy, J., Sabo, P.J. et al. High-throughput localization of functional elements by quantitative chromatin profiling. Nat Methods 1(3), 219-25 (2004).