Empty detailed.tsv

[rdmtinez]

I was executing some SMETANA runs and for every single one I noticed that there were no scores in the 'detailed.tsv'.... only the column headings. The GEMs I used were the ones output by CarveMe; both gap_filled and unfilled versions produced the blank detailed file.

[cdanielmachado]

What about the file called 'global.tsv', did it produce the expected result? What is the global SMETANA score for your communities ?

[rdmtinez]

global.tsv did produce the expected output, but the global SMETANA scores were 0.0

[cdanielmachado]

What were the values of MIP and MRO ?

[rdmtinez]

For that specific community it was 1 and .7317073...

[cdanielmachado]

Ok. I see what is happening. Did you specify a growth medium?

If not, then you are running simulations on a "complete" medium. Since each species can take whatever they need from the medium, the smetana score will be zero.

Only when at least one species depends on the others to survive, the smetana score will be larger than zero, and the cross-feeding interactions that mediate the dependency will be listed in the "detailed.tsv" file.

[rdmtinez]

Thank you so much! I looked at the code, but I couldn't figure out what was going on. :)

[cdanielmachado]

No problem :)

[rdmtinez]

Greetings Daniel,

Its been a while since I tinkered with SMETANA, but once again I am having a similar issue and was hoping you'd help me out.

To reiterate: I have made several hundred models using CarveMe as follow: A. carve --dna assembly.fna --fbc2 -o output_dir #default output, no gapfill, no predefined media B. carve --dna assembly.fna --init M9 --fbc2 -o output_dir #no gapfill, predefined media C. carve --dna assembly.fna -g M9 -i M9 --fbc2 -o output_dir #gapfilled on predefined media

For types B & C I have used three different media compositions (M9, artificial root exudates(ARE) and M9+ARE)

The models of type A all grow because they have unlimited media.

ALL of the models of the type B do not grow as per cobrapy's FBA; perhaps we have bad assemblies? The output scores for these look like Example 2, below.

The C models grow as they have been extensively gapped-filled. So I am using these with SMETANA currently. (When I contacted you back in October, I was using the (A) type models and your answer above still holds true for those), but now I'm not sure what's going with the models C....

You explained that "Only when at least one species depends on the others to survive, the smetana score will be larger than zero, and the cross-feeding interactions that mediate the dependency will be listed in the "detailed.tsv" file."---I am getting such results in some detailed.tsv, because as you say, there is likely some dependecy (see Example 1 below).

But I am still not getting any of the other scores MIP, etc. in global.tsv:

Example1: This one follows what you said, if there is growth dependency, then detailed.tsv is populated... why then are the MIP, NIP, and MRO values failing to compute? 
This is the case for every community that shows dependency (i.e. has a populated detailed.tsv file)--only the global smetana score is populated.

LjNodule213_LjRoot139_global.tsv 
community | medium | size | mip | nip | mro | smetana
-- | -- | -- | -- | -- | -- | --
all | M9ARE | 2 | Failed | Failed | Failed | 3.0

LjNodule213_LjRoot139_detailed.tsv
community | medium | receiver | donor | compound | scs | mus | mps | smetana
-- | -- | -- | -- | -- | -- | -- | -- | --
all | M9ARE | LjNodule213 | LjRoot139 | M_dtmp_e | 1.0 | 0.09 | 1 | 0.09
all | M9ARE | LjNodule213 | LjRoot139 | M_fol_e | 1.0 | 0.06 | 1 | 0.06
all | M9ARE | LjNodule213 | LjRoot139 | M_fum_e | 1.0 | 0.01 | 1 | 0.01
all | M9ARE | LjNodule213 | LjRoot139 | M_gmp_e | 1.0 | 0.39 | 1 | 0.39
all | M9ARE | LjNodule213 | LjRoot139 | M_h2s_e | 1.0 | 0.04 | 1 | 0.04
.....

============================================================
Example 2: In the next two examples,  I'm assuming that there is no growth dependecy, and so detailed.tsv is not populated, nor can the global.tsv scores be computed...

LjNodule213_LjRoot100_global.tsv
  | community | medium | size | mip | nip | mro | smetana
-- | -- | -- | -- | -- | -- | -- | --
all | M9ARE | 2 | Failed | Failed | Failed | Failed

 LjNodule213_LjRoot100_detailed.tsv
community | medium | receiver | donor | compound | scs | mus | mps | smetana
-- | -- | -- | -- | -- | -- | -- | -- | --

===========================================================
Example 3: However, in this one, the is a global.tsv 'smetana' score, but in Example 2, it fails to compute. This very small difference makes me believe there is a bug somewhere, and it maybe that is also causing the other scores to fail as well.

LjNodule214_LjRoot33_global.tsv 
  | community | medium | size | mip | nip | mro | smetana
-- | -- | -- | -- | -- | -- | -- | --
all | M9ARE | 2 | Failed | Failed | Failed |  

LjNodule214_LjRoot33_detailed.tsv
| community | medium | receiver | donor | compound | scs | mus | mps | smetana
-- | -- | -- | -- | -- | -- | -- | -- | -- | --