Frequently Asked Questions
Why do the Fold Change values in the differential gene expression analysis not correspond to the sample/control ratio of the FPKM values?
Answer: For gene expression analysis, the system actually uses a combination of the two packages Cufflinks and DESeq2. Fold Change and p-value are calculated according to the statistical model used in DESeq2 - all details on this are provided in the DESeq2 documentation. On the other hand, Cufflinks-based FPKM values are popular because they allow direct comparisons between different samples as well as between different genes in the same sample. Thus, these are additionally given for the user's orientation.
Is there a cutoff for the minimum raw counts prior to FPKM normalization?
Answer: The differential gene expression analysis is mainly based on the DESeq2 package which uses an appropriate model to account for low-coverage genes - please refer to the documentation. The counts behind the FPKM values are not filtered at all. For low-coverage genes, in particular, the user should therefore primarily rely on the DESeq2 fold changes while the FPKM values are rather to be understood as a supporting information.
What data transformation was used to build the heatmap?
Answer: After log2-transformation of the FPKM values, the genes are filtered according to the variance of their FPKM values across samples. Only the 1000 genes with highest variance are displayed.
Are the Fold Change values in the excel tables log2 or linear values?
The Fold Changes are linear. For any value smaller than 1 (i.e. for downregulation), the value is replaced by its negative reciprocal value. For instance, 2-fold downregulation is indicated by a value of -2 instead of 0.5.
Are the Fold Change values calculated independently for each of the comparisons? Or are all FoldChange values relative to all four conditions?
Answer: The Fold Changes and p-values are calculated independently, thus, there is no link between the calculations of for different comparisons.
Is it possible to retrieve the bam files, or the raw count tables, from the pipeline?
Answer: Both data sets do exist, but are not part of the standard content in the QuickNGS web interface. Please contact the QuickNGS operators to have them added into your account.
How can I use older versions of the Ensembl database instead of updating to the most recent version?
Answer: Use the script
scripts/public/download_ensembl.sh to download a particular version of the Ensembl database for your species. Go to the Ensembl subfolder of the
big data directory and relink
use to the directory containing the version you wish to use:
cd <datadir>/public/ensembl rm use ln -s 66 use
The system will then use version 66 for the next analysis. Please do not forget to re-set the link to
current after the analysis has started. This will avoid that an outdated version is being used in your following work.
Is the whole-genome resequencing pipeline capable of doing cancer genomics analyses?
Answer: There are a lot of peculiarities to consider when doing such an analysis on tumor-normal sample pairs. In particular, tumor samples typically contain a certain amount of stromal contamination which actually makes the sequencing data a mix of tumor and normal samples. That being said, the pipeline should be used only with care unless there is an obviously very low degree of contamination.
Previous topic: Troubleshooting
June 6th, 2017: Our paper on the cancer genome analysis platform QuickNGS Cancer has been published in Human Mutation.
September 30th, 2016: Please access the 'Multi-Layer Integration' area to combine RNA-Seq and ChIP-Seq data with an early release of our new multi-OMICS data integration platform.
June 23rd, 2016: Tools for gene set enrichment analysis using GO terms and KEGG pathways have been adopted into QuickNGS from version 1.2.2 on.
February 15th, 2016: The latest QuickNGS release now includes QuickNGS Cancer, a new platform specifically designed for cancer genome analysis.
January 4th, 2016: Bluebee High-Performance Genomics B.V., Delft, The Netherlands, have adopted QuickNGS into their cloud-based NGS analysis solution.
September 23rd, 2015: Please refer to our new FAQ to handle common problems with the QuickNGS results.
July 31th, 2015: The QuickNGS paper was published in BMC Genomics! Please cite this paper for all analyses based on the QuickNGS system.
July 11th, 2014: The first public version of the QuickNGS source code has just been released! Please click here to download the software.