search

sqjin / CellChat

R toolkit for inference, visualization and analysis of cell-cell communication from single-cell data
GNU General Public License v3.0
607 stars 140 forks source link

cell chat to analyze digital spatial profiling data #527

Open acjordan333 opened 1 year ago

acjordan333 commented 1 year ago

Hello Dr. Jin,

Thanks so much for creating such an amazing package. I'm a member of a lab analyzing digital spatial profiling data (from the NanoStringGeoMx platform) from human pancreas cancer specimens and we are wondering whether CellChat could be used to infer paracrine/autocrine signaling within specimens. The data consists of a count matrix with one row for each gene and each column representing an area of interest from the experiment. Each box has counts per gene from each area of interest which is composed of fifty to hundreds of cells. Therefore, our data is similar to single cell data in structure, but closer to a mini-bulk RNA dataset. Do you think that CellChat could predict signaling relationships between areas of interest, despite the fact that we would not be working with expression counts from single cells, but from hundreds of cells per area of interest? I'm happy to clarify further. Thanks for your help.

Best, Alex

sqjin commented 1 year ago

@acjordan333 Hi Alex, happy new year! Based on my understanding, your data is similar to 10X visium data where each spot contains multiple cells. Our new tutorial could be used to infer signaling between spots.

acjordan333 commented 1 year ago

Thank you Dr. Jin for the suggestion. I will look at the tutorial at once!

I had another question about what we can do if there are too few spots for analysis. For example, we have paired data between a few patients with approximately 4-6 spots per condition (pre vs. on treatment). CellChat appears to be telling us that that is too small a sample size to run the algorithm. In that case, could we duplicate the spots by multiplying them by some integer to try and run the software to get a sense of what autocrine/paracrine signaling is occurring in those samples (with the understanding that we would make it VERY clear we were doing this), or would that lead to false calling of signaling relationships in those samples? Thank you again for your time.

sqjin commented 1 year ago

@acjordan333 Can you please point out the function that gives the warning on the small sample size? You should be able to adjust some parameters to run the algorithm.