Low Fraction Gene Expression Reads in Spots in Visium 10x Fresh Frozen experiment

Question

[Sergio]

Hello,

I would like some expert advices about a problem that I encountered.

I've been analyzing some samples from Visium 10x FF, and I aligned my fastqs with spaceranger.

First, I aligned my histological images manually with Loupe, and I provided the .json file to spaceranger.

What happened is that I obtained this warning message in spaceranger web summary:

These are the other outputs of the summary:

Then, the first test that I did was the automatic alignment instead of the manual one, providing directly the .tiff histological image, without treating it with Loupe and without creating any .json file. By doing this, the Fraction Gene Expression Reads in Spots incresed a bit, but was still low:

Sequencing and Mapping section of the summary did not change at all from manual to automatic alignment (as expected, by the way), while this is the Spot section with the automatic alignment:

Of note, the same thing happened with other 3 samples, and those other 3 samples had a higher percentage of reads mapped to the genome (about 80%). Just to say that I do not think that can be the source of the problem.

What I thought to do, from the computational point of view, is:

• Check for eventual contamination of the samples, that could bring to high amount of reads outside the spots (how would you do that, by the way? by using kraken2?);
• Check the levels of the ambient RNA, that could also be the source of the problem, as suggested by spaceranger itself (how to do that? Would SoupX work, for example?).

Does anyone have a better idea? Do you think I am proceeding in the right direction?

Thank you very much for any suggestion!

[Trivas]

those other 3 samples had a higher percentage of reads mapped to the genome (about 80%)

Well I'd hope so. The issue is either problems with permeabilization (over permeabilization would increase your ambient RNA/background signal) or potentially the orientation of your histological images are mismatched with the orientation of your vizium slide. You should be able to still visualize your data over your tissue section and see very quickly if either of these are true.

[Sergio]

Hello Trivas,

Thanks a lot for your answer.

However, it is not clear to me how to understand if there is over permeabilization just looking at my data over the tissue. Also, about the orientation of the histological image: the parameter --reorient-images is set on true, so in principle the algorithm tries all the possible orientations and ends up with the best one.

Could you clarify better how you would perform your tests?

Thanks!