Is there a way to compare single-cell RNA seq and Ribotag-seq?

Question

[John]

Hello there,

I have two different dataset(different library), one is single cell RNA sequencing of 60 cells. And other one is 10 Ribotag IP sequencing (traslatome profiling). Is there any way to compare and correlate these two datasets?

When I try to plot PCA they cluster at different places but both datasets are belong to same cell types.

Thanks Jon

[Devon Ryan]

You'll have to think about what you really want to get out of the data. Most likely you'll want to look at how DE genes in your Ribotag data change in different cells in your scRNA-seq data (with only 60 cells, I expect you explicitly took different cell types).

[John]

Hi Devon,

I plotted the cell trajectory using monocle. When I tried to plot, I used different different datasets of single cell seq and my ribotag seq data. I used normalizebetweenarrays function from limma, to normalize the datasets, to have same column sums. All the single cells are looking good, that they are staying in acceptable positions in the trajectory, but my ribotag seq data stays at the end of trajectory, but they must be in the middle of trajectory(based on their cell maturation level).

And how can I do this

how DE genes in your Ribotag data change in different cells in your scRNA-seq data

[Devon Ryan]

Combining those two datasets in a trajectory won't produce sensible results.

[John]

Okay, as the single cell datasets have the information of maturation. Can i show particular ribotag-seq belongs to particular maturation level? Is it possible?

[Devon Ryan]

That's possible but quite difficult. Your setup doesn't lend itself to directly comparing the two types of datasets. What you might be able to do is look at genes changing around the time point you want in the trajectory and then look at how they're expressed in your Ribotag data. Reviewers aren't going to have plenty of objections to that, but if the results fit in with a broader story that sort of procedure should work.

[John]

okay, Can you tell me some examples/R packages/algorithms to do.

Thanks

[Devon Ryan]

You'll have to invent this.

[John]

That's exciting, thanks for your discussions.

[John]

Hi Devon, I got Dynamically regulated genes across the maturation trajectory(for single cell data), around 1000 genes, with 24 pseudo-time points.

I have calculated correlation (spearman's/Pearson) between each ribotag data and all the pseudotime points. (to find which pseudo time expression pattern is similar to ribotag data)

All the ribotag data highly correlated with more mature state. but in reality, they are not.

Is it a right way of doing? do you have any suggestions?

[Devon Ryan]

That's probably as good as can be done. You might have to tweak filtering to get things to align a bit better with your background knowledge of how things should look.