Important notes to the user of this service
- For reliable transcript expressions, a sufficiently high coverage (at least 50 million reads) and a considerable amount of samples (>5) for each class is recommended.
- If your data is not yet
aligned to a reference genome and available in .bam format, our
Read-Mapping service will provide adequate input data.
Required input data
- sequence reads aligned against a reference genome (e.g. hg19 for homo sapiens) in .bam format.
- assignment of samples to 2 or more named classes (e.g. in a table)
- list of genes expressed differentially between given classes
- list of transcripts expressed differentially between given classes
- various summarizing plots (histogram/boxplot of expression distribution in each class, bi-clustered expression heatmap of differentially expressed genes/transcripts against the given classes)
Additional options on request
- bias correction
- upper-quartile normalization to improve robustness for input data with low coverage
- transcript assembly (planned)
Methods & Implementation
Estimation of gene/transcript expression, normalization and detection of differentially expressed genes/transcripts is conducted with cufflinks 1.3.0
. Genes and transcripts with a (multiple testing corrected) p-value at most 0.05 are selected as differentially expressed. Summarizing plots are generated with the R package cummeRbund 1.1.3