A question about FBA and FVA

[Erique29]

Hi , After constructing metabolic models using this wonderful tool, I want to validate the accuracy of my models using BIOLOG plates, like the paper 'The functional repertoire contained within the native microbiota of the model nematode Caenorhabditis elegans'. I used the data 'MYB11 and MYB71' of this published study to test. I referred to your method in Lactic acid bacteria in yogurt production, but have not find the right result of carbon sources and by-products in Fig. S9. Could you help me with this? THANKS!

[jotech]

Hi @Erique29, great idea to validate the models using phenotypic data! If you already have the biolog data available, I would recommend to use a check for electron donor instead of standard FBA with biomass optimization. It's more close to the actual biolog experiment to test the capacity of a compound to serve as electron source for electron carriers like NAD or quinones. Some example code for this can be found in the code repository of the gapseq paper. Do you think this could be of any help in your case? In the past, I found it not always easy to discriminate between growth/non-growth using biolog, a foldchange > 1 of OD corrected values (early vs. late) worked best if I remember correctly. What is your approach here? Thanks for your feedback!

[Erique29]

Thanks @jotech ! This is indeed a good way to validate the BIOLOG data using the electron transfer which works the same as BIOLOG. Unfortunately, I do not have BIOLOG data now. I will start the BIOLOG experiment in next few weeks. And I will communicate with you about the phenomenon you have found. By the way, I would like to validate the models using both the electron transfer and the FBA to have a mutual confirmation. And I'm also interested in the carbon sources and metabolic byproducts. How can I modify the method I used above to get the similar results as Fig. S9? It actually troubled me for many days.

[jotech]

Hi @Erique29, the potential carbon sources and byproducts from Figure S9 were determined by flux variability analysis (FVA). If there was a non-zero flux of the respective exchange reaction while guaranteeing at least 10% of the maximum growth yield, then a compound was assumed to be a carbon source (flux<0) or byproduct (flux>0). Does it makes sense to you?

[Erique29]

Hi, it does make sense to me. I will try it! One more question, I tried MEMOTE to score my model, but I found the annotation-genes were all zero. And the total score was 77. Is there any problems in my model? Thanks! index.zip

[Waschina]

Hi! No, this is nothing to worry about. The reason is that gapseq follows a unique approach to gene-identfiers in model. The gene names in gapseq indicate the contig-ID and the position of genes in the input genome sequence (see for instance also at the discussion in the issue #71).

[Erique29]

Yeah, It explains the matter. Thanks for your help!

[jotech]

please reopen if there are still things unclear :)

[Erique29]

Hi @jotech, I finished the BIOGLOG experiment this week, and I found that the foldchange > 1 of OD590 of some wells were not easy to discriminate. Because they were lavender like some wells with foldchange < 1. According to my observation, foldchange >2 of OD590 is better to discriminate the growth. Because the color of the wells (foldchange >2) were all dark purple after 48 h. At last, I used BIOLOG GEN III, and my initial OD590 of bacterial suspension was 0.05.

[jotech]

interesting, thank you for reporting your experiences here!