Dear colleagues,
I was reading the paper "Integrated biomarker response: a useful tool for ecological risk assessement", by Beliaeff et al 2002, and I was trying out the computations as in the example (Computation of IBR at station W3 (Warnemunde Estuary) in March 1995, p1318) and there's one thing I couldn't quite follow.
When the authors write about the minimum value (Min) for all stations and/or surveys for each biomarker, I would think the minimum value (that later on we add to Z to calculate the score S) would be obtained directly from the original dataset (not shown in the article), but what I don't understand is how the Min values are below zero for biomarker levels. I mean, measured enzyme activities are always above zero.
Therefore my questions are: i) how is it we obtain the Min value? or iii) are the minimum values relative values, such as a fold-change in enzyme activity?
In addition, when we have to place a + or - sign in Z (showing upregulation or downregulation in biomarker levels), does that correspond to what is expected in theory or to what was actually observed in the data?
Thanks a lot.
Kind regards
http://onlinelibrary.wiley.com/doi/10.1002/etc.5620210629/abstract
I can send you reference of more two papers where some details are available in the supplementary data. Hope that works. If not you can just contact Nuno Ferreira (the first author) and ask him for details.
Susana
Hello Carolina!
I'm not really experienced with IBR, but if I understood your questions, these are my suggestions:
I) The min value might be obtained after the Z transformation. Therefore, it will be necessarily negative, and when you add it to all values, the distribution of your data become =0 or positive.
II) Yes, in this case minimum value will be relative to average of all values and SD.
III) + and - might be decided as understood in theory as positive of negative response to the pollution. E.g. a positive relationship (+) with LPO, but inhibition of antioxidant responses (-). This is the tricky part because several biomarkers present a biphasic response, which is a bit harder to implement on IBR. For instance when low levels of contamination increase an specific antioxidant response, but high levels of pollution deplete it.
Hope it helps.
All the Best
Abel
This is indeed the tricky thing with this IBR as you have to start with your own assumptions. And you need to decide if the "bell shaped" enzymes will respond one way or the other (going up or down) when exposure is present.
Susana
Dear Dr Susana and Dr Abel
Thank you so much for your answers, they were really helpful and gave me insight about further questions I hadn't considered before.
I do have biomarkers with biphasic response, and now I'm thinking the best way to solve that - and make sure the IBR is as much representative of the original data as it can be - is by plotting the results for each sampling time (so let's say at day 0, 7, 14, Z will be positive (+) and at day 21 and 28 Z will be negative (-) for that same biomarker, for example, in order to represent that change in activity/inhibition).
In my research I'm not studying chemical pollution, but rather variation in physical stressors influenced by climate change, on marine species. As I have many species, target tissues, biomarkers and stress categories, I'm thinking the best method to integrate these multilayered data are IBRs. In your professional opinion, since the objective is to be able to represent a multivariate dataset in a graph, are IBRs the best solution? Or are you familar with any other methods?
I know IBRs were originally developed for chemical pollution analyses, but since the concept is the same, I think they can be a pretty good solution - although I'm asking because I don't think I've seen them being used in climate change studies.
Thank you!
All the best,
Carolina
Hello Carolina,
I would say in theory IBR could integrate other stressors than chemical. Also, biphasic response is a problem common to diverse physic-chemical-biological pollutants.
Personally, instead transforming each biphasic biomarker into two variables with positive and negative values, I would use the square of the diference from control z score. In this sense it doesn't mater which direction the change is, it will be penalized equally as it goes a way from control conditions (that can be in time and or space). This way is lees subjective than you deciding when/if the response was biphasic and whether was in its inhibition or stimulation phase.
One alternative to IBR that I came across some years ago was the Expert system, which already accounts for biphasic responses (Dagnino, A., Allen, J.I., More, M.N., Broeg, K., Canesi, L., Viarengo, A. 2007. Development of an expert system for the integration of biomarker responses in mussels into an animal health index. Biomarkers. 12:155-172). it seems very promising and mathematically more robust than IBR. However for there is less bibliography and the authors were not really helpful when I tried to contact them.
Good look with the calculations, and let me know about the outcomes. :-)
All the Best
Abel
Thanks a lot Abel! Hopefully I will soon have some interesting results to publish!
Hello I have a question about the computation of IBR. I am working on gene expression and my data is not homogeneous because of I am comparing different species from different ecotypes. So, I have some extreme values for some genes and because of in IBR we use min value at a point, those values totally over-shades other genes and time points. To smooth the response I transformed my data with LOG and LN(x+1) but this time I obtain negative scores and negative IBR values. Can someone explain me where am I doing wrong?
Regards
Anyone knows if is available a software, excel or a R package to integrate data from each biomarker? For example, for BIES by Sforzini et al. 2011, there is a software to integrate each data very easy.
Dear colleagues; This subject interests me too.
Here are two articles that can help you. kind regards
- Badges
- Science method
- Similar topics
- Biological Science
- Ecology