Based on the criticism by some coauthors of our recent manuscript the decision has been made that the work needs a major revision. What should be that revision? Some key points are listed here.
General questions
More data and/or better model?
Explaining variation of allelic bias
- Given the performed model checking and the estimated significance (CI, p-values) how solid are the results?
- if the employed model is correct the interpretation of CI and p-values is clear, isn’t it?
- in the other case: what is specifically incorrect about the model and to what extent does it affect the results?
- What are the most efficient ways to make results more solid
- more data: requires reimplementation of low-level pipeline and has little use if the model is incorrect
- better model: requires adapting an existing one (e.g. limma/voom) or designing a new one
Genome-wide assessment: what’s new?
- what are the most important findings considering two points:
- several studies, and even Dulac’s own group, have shown that Dulac et al previous estimate of >1000 imprinted genes is hugely inflated
- the Lappalainen group (Baran et al) already assessed allelic bias in the human brain albeit using a smaller data set of not specified brain regions
Specific collaborators, coauthors
CMC WorkGroup
I must become a member in order to access data and results. Who can grant me admission?
Gabriel Hoffman
Thoughtful comments. The suggested analyses and modifications sound useful but would require reimplementation of low-level pipeline. The suggested improvements to the presentation would be also useful and they are easy to do.
- Are the regression results solid? If not: more data and/or better model? See above.
- terminology
- “allelic bias”: OK or replace with something else (e.g. allelic imbalance, allele-specific expression)?
- suggestions for replacing wnlm.Q, logi.S, etc