|
Status |
Public on Jun 06, 2012 |
Title |
Whole transcriptome diff. expression analysis of HFF cells using TileShuffle variant B (G0 vs. G1)_chipE |
Sample type |
RNA |
|
|
Source name |
Human foreskin fibroblasts (HFF) residing in G0 phase compared to HFF cells residing in G1 phase of mitotic cell cycle
|
Organism |
Homo sapiens |
Characteristics |
cell line: HFF data processing: TileShuffle
|
Extracted molecule |
total RNA |
Extraction protocol |
Trizol extraction of total RNA was performed according to the manufacturer's instructions.
|
Label |
biotin
|
Label protocol |
10µg RNA were analyzed on Affymetrix Human Tiling 1.0 array set according to the manufacturer's protocol.
|
|
|
Hybridization protocol |
Affymetrix Human Tiling 1.0 arrays were proceed using the GCS3000 7G system according to manufacturer's protocol.
|
Scan protocol |
Affymetrix Human Tiling 1.0 array were proceed using the GCS3000 7G system according to manufacturer's protocol.
|
Description |
Whole transcriptome differential expression analysis of HFF cells on Affymetrix Human Tiling 1.0 array set. Cells were synchronised by serum starvation and transcriptome-wide changes occurring at the transition from G0 phase to G1 phase detected. Expression data and fold changes were processed with our new permutation approach TileShuffle.
|
Data processing |
Affymetrix Human Tiling 1.0 probes were mapped to human genome assembly hg18 using the corresponding BPMAP files. The expression data were processed with our new permutation approach TileShuffle. TileShuffle was applied using only perfect match (PM) probes, a window size of 200, the arithmetic mean trimmed by maximal and minimal value as scoring function, three GC content bins, 10000 permutations, and a q-value threshold of 0.05. In order to estimate the significance of a window score we repeatedly permute all probe intensities across the array while interchanging only those that belong to the same sequence-specific affinity bin, recompute the window scores, and compare them with the original ones. By counting the number of permuted windows with higher score, we estimate empirical p-values of windows, and report q-values adjusted according to Benjamini and Hochberg ("Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing". Journal of the Royal Statistical Society. Series B (Methodological), 57:289-300, 1995.). Analysis of differential expression was performed with the same parameters, except of 100000 permutations, aiming at accommodating the more rugged nature of expression difference signal. Here we use variant B to detect differentially expressed segments, i.e. it is assumed that entire windows are either up- or downregulated between two conditions and converse behaviour of neighboring probes is a consequence of non-specific hybridization. In order to correct for this bias, the presumed direction of regulation is initially assigned to each window on the basis of the sign of its expression scores. Subsequently, all converse probes, i.e. probes with negative log-fold change within positive windows or vice versa, are ignored and neither permuted nor incorporated into the score calculation for differential expression. Consequently, positive and negative windows are compared to different background distributions. To asses the significance of a window score a one-tailed empirical p-value is estimated (according to the corresponding background distribution) and corrected for multiple testing, similar to the one-state analysis. Here we use a q-value < 0.1 which results in an FDR of 18%. To avoid that signal intensity variation at the detection limit is classified as differential expression, we require that differentially expressed intervals must also be significantly expressed in at least one of the investigated conditions, and call these intervals highdiff. Following this definition, TileShuffle takes regions found to be significantly high expressed in at least one of the compared states as input for the two-state analysis, and assesses differential expression solely on the expressed segments.
|
|
|
Submission date |
Feb 29, 2012 |
Last update date |
Jun 07, 2012 |
Contact name |
Kristin Reiche |
E-mail(s) |
kristin.reiche@izi.fraunhofer.de
|
Organization name |
Fraunhofer Institute for Cell Therapy and Immunology
|
Department |
Diagnostics
|
Street address |
Perlickstr. 1
|
City |
Leipzig |
ZIP/Postal code |
04103 |
Country |
Germany |
|
|
Platform ID |
GPL6160 |
Series (2) |
GSE36164 |
Differential expression analysis of HFF cells in G0 phase compared to cells in G1 [TileShuffle] |
GSE36190 |
Differential expression analysis of HFF cells in G0 phase compared to cells in G1 using TileShuffle |
|
Relations |
Reanalysis of |
GSM739233 |
Reanalysis of |
GSM739247 |