Error ID 11 - Failed to estimate expression variance

Error Description

You received this error because you tried running a differential expression pipeline, but there is too little expression variation in your samples. There are several potential causes for this: (1) the number of reads mapping to your genes is too low so that expression estimation is too unreliable, or (2) their truly is very little biological variation among your samples.

Solutions

Solution 1

Double check that you assigned your samples to the appropriate groups. 

Solution 2

Double check the pipeline that aligned your reads and estimated gene expression (e.g. the "Expression count (STAR)" pipeline). At the bottom of the analysis page you should see a table of per-gene expression estimates (see the example image). Look at the Count column, which shows the number of reads mapping to each gene. If you see the vast majority of genes with counts of 10 or less (for example), then you might not have enough reads in the sample to get reliable expression estimates. If this is the case you have several options:

1. Re-sequence your samples at a higher sequencing depth.
2. Rerun the pipeline and have it include multi-mapped reads in expression estimation (though this is not advisable).
3. If you see many of your reads are being removed in the trimming step, this could indicate your  sample is of poor quality. Nonetheless, you can rerun the expression count pipeline and slightly decrease the stringency of the trimming threshold parameters. However, this is also not very advisable.

Solution 3 - contact us

If you still cannot resolve your issue, please don't hesitate to contact us by either:

• Creating a ticket here: http://support.basepairtech.com/support/tickets/new
• Message us using the chat icon in the lower right corner of your screen on the Basepair dashboard.

Further information

If you ran either the  "Differential Expression (DESeq, compare 2 groups)" or "Differential Expression (DESeq, 3+ group comparison)" pipelines, the underlying software (DESeq2) does offer some more alternatives for providing variance estimates. However, Basepair currently does not support those other options, but you read about them on the DESeq2 vignette: https://bioconductor.org/packages/release/bioc/html/DESeq2.html