Survival Analysis Cox Regression

[maurertom]

Report

Dear ICBI Team, Dear Gregor!

Thank you so much for taking time to develop this nice tool :). I would also really like to use it in my projects!

I got a question: Is the Cox regression for survival analysis allready integrated in this Version of shears ? From what I see there is only logistic regression in def_test_cell() ?

Or am I just using it wrong?

I tried it like this - would this implementation work?

def _test_cell_cox(formula, df, duration_col, event_col):
    from lifelines import CoxPHFitter
    
    cph = CoxPHFitter()
    cph.fit(df, duration_col=duration_col, event_col=event_col, formula=formula)
    
    cell_weight_coef = cph.params_['cell_weight']
    cell_weight_pval = cph.summary.loc['cell_weight', 'p']
    
    return cell_weight_pval, cell_weight_coef

Thanks for your time and hope the question is not to stupid,
Have a nice day
Thomas

Versions

[vjmarteau]

Hi Thomas,

you can find the cox implementation of shears in this pull request. Gregor is currently on holiday and will be offline for a month, but I think the implementation will not change much once it is merged. This version of shears is also what we have been using in our manuscript.

I will try to update the shears tutorial in the next months, in the meantime you can install shears using pip install git+https://github.com/vjmarteau/shears.git and run it as follows:

covariate_str = (
    "C(stage, Treatment(reference='early'))"
    " + C(sex, Treatment(reference='male'))"
    " + age_scaled"
    " + C(project, Treatment(reference='TCGA'))"
)

sh.tl.shears_cox(
    adata_sc,
    adata_bulk,
    duration_col="OS_time",
    event_col="OS_status",
    covariate_str=covariate_str,
    n_jobs=-1,
)

sh.tl.shears_glm(
    adata_sc,
    adata_bulk,
    dep_var="microsatellite_status",
    covariate_str=covariate_str,
    n_jobs=-1,
)

Let me know if you have any problems, questions or suggestions.

Best
Valentin