Closed sgosline closed 2 years ago
I'm updating the manuscript outline based on our current tests and discussions.
2- Results - extensible software framework enables facile comparison of proteomics-based deconvolution algorithms Figure 1 - framework diagram and example of description of overall, correlation, and subtype matrix.
Song
3-Results - Results on simulated data show that.... Francesca/Sara Figure 2
4- Results - Cell-type correlations confer that MCP counter is most robust between mRNA and protein comparisons Figure 3 Sara
5- Results -Immune subtype comparisons show that .... Pietro Figure 4
6 - Results - Imputation has little impact on strong-performing deconvolution algorithms Song Figure 5
7- Discussion - Value of protein deconvolution over mRNA This is a key point that we need to drive home
8- Discussion - lack of gold standard and need for more validation the rieckmann data is good but need tumor sample and cell mixing experiment
9- Discussion - ability to plug in additional algorithms/tools
Also need to add tumor-normal comparison in supplemental.
Manuscript is now in process on google dirve.
I think it would be helpful to allocate specific parts of the manuscript so that we can be focused about task allocation and what our ultimate goals are (and are not!).
1- Introduction
2- Results - benchmark software framework Song
3- Results - comparison of algorithms (mRNA to protein) across single matrix by correlation across patients - which cell types are most similar??? Song
4- Results - comparison across matrices (mRNA to protein) across single algorithm - which matrix works best for both mRNA and protein? song
5-Results - comparisons across tumors (mRNA to protein by patient) across matrix by correlation across cell types - which cancer types are the most similar? Song
6 - Results - Impact of protein imputation on deconvolution Anna
7- Results - flexibility of algorithms - this could be discussion Francesca
8- Results - comparison with iAtlas immune subtypes - inferred on CPTAC data. Pietro