feat: gene essentiality workflow

[mihai-sysbio]

Main improvements in this PR:

This PR aims to automatise the gene essentiality checks as introduced in #563 with a GitHub workflow, also discussed in that PR thread.

I hereby confirm that I have:

• Tested my code on my own computer for running the model
• Selected develop as a target branch
• Any removed reactions and metabolites have been moved to the corresponding deprecated identifier lists

[mihai-sysbio]

@haowang-bioinfo the output of this line is very hard to capture/use. Do you have any suggestions to make this output more compact?

[mihai-sysbio]

As seen in this workflow run, the output is way too rich to fit in a PR comment. @haowang-bioinfo how would you suggest to change it to end up with a nice summary like here?

[haowang-bioinfo]

no, I've never thought about this. Because they are non-binary indicators that are hard to be used for evaluating purpose in actions

[mihai-sysbio]

Also noting here this relevant research article Synthetic lethality in large-scale integrated metabolic and regulatory network models of human cells.

[haowang-bioinfo]

Also noting here this relevant research article Synthetic lethality in large-scale integrated metabolic and regulatory network models of human cells.

yes, it mentioned another method gMCStool that can predict essential genes in context-specific models.

Other checks than essentiality are also welcome

[feiranl]

@mihai-sysbio Could you explain the reason for the failing task?

[mihai-sysbio]

@mihai-sysbio Could you explain the reason for the failing task?

The task isn't actually failing, really. As mentioned in the comment above, it's simply that the output of the gene essentiality is so long that it cannot be posted as a PR comment. I couldn't figure out how @haowang-bioinfo ran the calculations to obtain a much more compressed output. As visible in the workflow changes, the code here is just calling functions defined elsewhere.

[haowang-bioinfo]

I couldn't figure out how @haowang-bioinfo ran the calculations to obtain a much more compressed output. As visible in the workflow changes, the code here is just calling functions defined elsewhere.

the concise output was an outcome of a series of manual operations of running functions, copy/paste results, and modifications with text editor

[mihai-sysbio]

If these last manual steps could be ran as a function, then we would be ready to run the gene essentiality analysis on every PR 💪🏻

[edkerk]

As is suggested for MACAW results in #840, a summary output could be automatically posted in the PR, while a more detailed text file could be written and committed as part of each PR. This could be organized in the same folder as the MACAW results, and a README.md file could indicate for both output files from which PR they derived from.

[github-actions]

This PR has been automatically tested with GH Actions. Here is the output of the MACAW test:

Starting dead-end test...
 - Found 1514 dead-end metabolites.
 - Found 1319 reactions incapable of sustaining steady-state fluxes in either direction due to these dead-ends.
 - Found 1977 reversible reactions that can only carry steady-state fluxes in a single direction due to dead-ends.
Starting duplicate test...
 - Skipping redox duplicates because no redox_pairs and/or proton_ids were provided.
 - Found 447 reactions that were some type of duplicate:
   - 0 were completely identical to at least one other reaction.
   - 13 involve the same metabolites but go in the opposite direction or have the opposite reversibility as at least one other reaction.
   - 447 involve the same metabolites but with different coefficients as at least one other reaction.

This and a more detailed output from MACAW are also committed to data/macawResults/.

Note: In the case of multiple test runs, this post will be edited.

[github-actions]

This PR has been automatically tested with GH Actions. Here is the output of the gene essentiality test:

     cellLine     TP     TN     FP    FN     accuracy    sensitivity    specificity       F1         MCC   
    __________    __    ____    __    ___    ________    ___________    ___________    ________    ________

    {'DLD1'  }    36    2185    59    279    0.86792       0.11429        0.97371       0.17561     0.15291
    {'GBM'   }    34    2165    61    298    0.85966       0.10241         0.9726       0.15925      0.1333
    {'HCT116'}    46    2207    53    309    0.86157       0.12958        0.97655       0.20264     0.19047
    {'HELA'  }    30    2263    69    254    0.87653       0.10563        0.97041       0.15666     0.12398
    {'RPE1'  }    14    2204    81    259    0.86708      0.051282        0.96455      0.076087    0.025853
    {'all'   }     7    2408    92    109    0.92317      0.060345         0.9632      0.065116      0.0254

Note: In the case of multiple test runs, this post will be edited.

[edkerk]

Finally 😄 it seems to work! Ready for review.

[mihai-sysbio]

This is amazing 🤩 thank you @edkerk for seeing this through 💪🏻

[mihai-sysbio]

As a contributor to this PR I cannot formally review it - only comments are allowed - so please consider this an approval 👍🏻