I have read some papers on omics science and listened different lectures on the the changes induced by a treatment or a disease in a model.
Most of the researchers use 1.5 or 2 as fold-change criteria to discriminate between "intra"-model variability and treatment/disease effects.
Unfortunately I can't find references supporting these criteria.
Can you help me? I need some references to compare our criteria for a different endpoint (LDH/DNA releases) with criteria adopted in other fields of life science.
Thank you in advance!
I don't think that there are references. I can not imagine any scientifically sensible reason to use a "2-fold cut-off" as a rule. This is just a "convenience rule", providing not too few and not too many "candidates" in typical omics-setups.
The impact of the *measured*(!) gene regulation depends on many things that are typically not considered in such studies. For example, does the gene play a central role in perturbing or regulating one ore more signalling cascades? Is the regulation a trigger of further events or is it the result of a re-adjustement of a steady state to accomodate for an altered environment? Is the biological metarial a well-defined cell-type and is the time point of the sampling/measurement relative to the stimulus and the cell-cycle similar for all the cells in the sample?
The "rule of 2-fold cut-off" treats transcriptions factors and kinases and cytokines and metabolic enzymes and modulators and co-factors and and structural proteins etc. all in the same way. Typically cell mixtures (homogenates) are analyzed, or cell cultures of unsynchronized cells where the stimulus hits every cell in a different stage of a cycle are used. It may be that only a minor fraction of the cells in the sample need to respond, and this may translate to a very small measured fold-change for the whole sample. Changes in cell composition in the samples may lead to very large changes that are not relevant for the stimulus.
To my optinioon, a more senible criterion for "candidate selection" is the false-discovery rate, although this also has its owns problems and draw-backs.
I don't think that there are references. I can not imagine any scientifically sensible reason to use a "2-fold cut-off" as a rule. This is just a "convenience rule", providing not too few and not too many "candidates" in typical omics-setups.
The impact of the *measured*(!) gene regulation depends on many things that are typically not considered in such studies. For example, does the gene play a central role in perturbing or regulating one ore more signalling cascades? Is the regulation a trigger of further events or is it the result of a re-adjustement of a steady state to accomodate for an altered environment? Is the biological metarial a well-defined cell-type and is the time point of the sampling/measurement relative to the stimulus and the cell-cycle similar for all the cells in the sample?
The "rule of 2-fold cut-off" treats transcriptions factors and kinases and cytokines and metabolic enzymes and modulators and co-factors and and structural proteins etc. all in the same way. Typically cell mixtures (homogenates) are analyzed, or cell cultures of unsynchronized cells where the stimulus hits every cell in a different stage of a cycle are used. It may be that only a minor fraction of the cells in the sample need to respond, and this may translate to a very small measured fold-change for the whole sample. Changes in cell composition in the samples may lead to very large changes that are not relevant for the stimulus.
To my optinioon, a more senible criterion for "candidate selection" is the false-discovery rate, although this also has its owns problems and draw-backs.
- Badges
- Science topic
- Similar topics
- Medicine
- Life Sciences