A problem about get_stat_df() function

[WeiQin2001]

Thank you very much for your work, very useful for identifying key genes between different differentiation trajectories!!!! But I found a small problem in my personal use, log2FC is supposed to be the expression change of the query group relative to the reference group, here np.mean(rgex)/np.mean(qgex) is the expression change of the reference group relative to the query group, so I guess there should be an error here?
I think

[dinithins]

Hi @WeiQin2001, Many thanks for highlighting this! You are correct, the standard l2fc calculation indeed takes the expression change of query group relative to the reference group. However in the current g2g version, it does the opposite and reports the expression change of the reference group relative to the query group. I will add a note about this in README so that the users are clear on interpreting the values. Hope to change this in a next release. Until then the users can take the negative of df['l2fc'] as the standard l2fc results.

[WeiQin2001]

Okay~ I have another question, when using the gene2gene algorithm, when there are multiple cell differentiation trajectories in the dataset, is it necessary to single out a particular differentiation trajectory for the sake of the interpretability of the results? For example, when analysing T-cell differentiation trajectories, is it necessary to separate out CD4T cell differentiation trajectories and CD8T cell differentiation trajectories for comparison between disease and control groups? | | 魏钦 | | ***@***.*** | ---- Replied Message ---- | From | Dinithi ***@***.***> | | Date | 8/10/2025 21:02 | | To | ***@***.***> | | Cc | ***@***.***> , ***@***.***> | | Subject | Re: [Teichlab/Genes2Genes] A problem about get_stat_df() function (Issue #17) | dinithins left a comment (Teichlab/Genes2Genes#17) Hi @WeiQin2001, Many thanks for highlighting this! You are correct, the standard l2fc calculation indeed takes the expression change of query group relative to the reference group. However in the current g2g version, it does the opposite and reports the expression change of the reference group relative to the query group. I will add a note about this in README so that the users are clear on interpreting the values. Hope to change this in a next release. Until then the users can take the negative of df['l2fc'] as the standard l2fc results. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: ***@***.***>

[dinithins]

Yes, this is because the Genes2Genes algorithm only supports pairwise trajectory alignment. If a gene expression distribution is coming from multiple differentiation lineages, it is a mixture of distributions which will confound true similarities and differences between the reference and query groups for each lineage context.

[WeiQin2001]

Ok, I get it, thank you so much for your reply and your work!!!! | | 魏钦 | | ***@***.*** | ---- Replied Message ---- | From | Dinithi ***@***.***> | | Date | 8/10/2025 21:24 | | To | ***@***.***> | | Cc | ***@***.***> , ***@***.***> | | Subject | Re: [Teichlab/Genes2Genes] A problem about get_stat_df() function (Issue #17) | dinithins left a comment (Teichlab/Genes2Genes#17) Yes, this is because the Genes2Genes algorithm only supports pairwise trajectory alignment. If a gene expression distribution is coming from multiple differentiation lineages, it is a mixture of distributions which will confound true similarities and differences between the reference and query groups for each lineage context. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: ***@***.***>