gapseq
gapseq copied to clipboard
jotech
loss of model attributes when converting SBML to RDS causing error in gapfilling
for bug reports and errors please report the output of: ./gapseq test Hi, Thank you for the software, I am finding it very useful. I added reactions in different compartments (extracellular, periplasm, etc) in the sbml file (xml) because I find it easier working on python instead of R. But I wanted to know if I can convert this file into RDS to make use of the gapfilling process. Do you think this is feasible?
Thank you!
Esteban.
hi, yes you can load xml files in R using the package sybilSBML
library(sybilSBML)
mod <- readSBMLmod("my.xml")
saveRDS(mod, "my.RDS")
Hi, thanks for the reply.
Sorry in my first comment didn't specify with details. I have done this but I have an error when I want to use the RDS file with gapseq. Doing with ecoli in toys:
gapseq fill -m ecoli-draft.RDS -n ecoli-medium.csv -g ecoli-rxnXgenes.RDS -c ecoli-rxnWeights.RDS
...
Final growth rate: 1.941537
Uptake at limit:
Maltose:2.5, NH3:10, N-Acetyl-D-glucosamine:5, Neu5Ac:5, Melitose:1.67, Sucrose:2.5, O2:12.5
Top 10 produced metabolites:
H+:50.658, Acetate:49.47, H2O:34.554, CO2:20.421, H2:2.635, EX cpd11416 c0:1.942, Galactose:1.67, 5-Methylthio-D-ribose:0.007
Warning message:
Model file already exists and will be overwritten!
Wrote file ecoli.xml
Patching file ecoli.xml ... done
Then I use ecoli.xml to read it and export in RDS (I am aware that gapseq generate the RDS, but just to demostrate in the case when I want to modify the xml and then use it to gapfill)
library(sybilSBML)
mod <- readSBMLmod("ecoli.xml")
> mod <- readSBMLmod("ecoli.xml")
reading SBML file ... OK
getting the model ... OK
creating S and parsing constraints ... OK
GPR mapping ... OK
cleaning up ... OK
validating object ... OK
> mod
model name:
number of compartments 3
c0
e0
p0
number of reactions: 3035
number of metabolites: 2451
number of unique genes: 1252
objective function: +1 EX_cpd11416_c0
saveRDS(mod, "ecoli2.RDS")
When I want to use ecoli2.RDS in gapseq I have the following error:
gapseq fill -m ecoli2.RDS -n ecoli-medium.csv -g ecoli-rxnXgenes.RDS -c ecoli-rxnWeights.RDS
Loading model files ecoli2.RDS
Error in `[.data.frame`(mod.orig@mod_attr, , "annotation") :
selezionate colonne non definite
Calls: grepl -> [ -> [.data.frame
Esecuzione interrotta
gapseq version: 1.2 315f35d
I attach the files generated just in case. Do you know if I can solve this issue?
Thanks ecoli.zip !
Hi, I see; thanks for the clarification! The problem is some data loss from importing and exporting the model again. I just committed a quick fix for this, but we should also take care of the data loss in general.
Reproducible example for the divergence in model attributes that could cause errors when a sbml is imported in R and used as input for gapfilling:
library(sybilSBML)
library(sybil)
mod.rds <- readRDS("toy/ecoli.RDS")
mod.sbml <- readSBMLmod("toy/ecoli.xml.gz")
mod.sbml@mod_attr
notes
1 <notes>\n <html xmlns="http://www.w3.org/1999/xhtml">\n <p>gapseq version: 1.1 4a17997</p>\n </html>\n</notes>
mod.rds@mod_attr
annotation
1 tax_domain:Bacteria
dim(mod.rds@react_attr)
[1] 3029 26
dim(mod.sbml@react_attr)
[1] 3029 1
dim(mod.sbml@met_attr)
[1] 2451 3
dim(mod.rds@met_attr)
[1] 2451 29
Hi jotech, Thank you for the reply. Now is clear why I can't do this. I modified my script to use it in R, so I add the reactions to the draft RDS file instead of the xml using addReact(). In this way now I can work, if you can find the solution to export RDS from xml would be useful also.
Thanks!
Esteban.
Hi,
I'm afraid, that we can't get the additional columns exported to sbml.
SBML has only limited and pre-defined columns of model, reaction and metabolite attributes that are exported. In the RDS file, we keep some additional information that is helpful for debugging and further analysis – information that gets lost when exporting to SBML.
Hi Silvio, Thank you for the reply and take care about this. I will work directly on the RDS files through the sybil package then. I will close this thread but the help was useful to know this issue I was having.
Thanks.