Mapping mixed bulk RNAseq reads

,
1

I’m using a model where I infused human cells into the organ of a different species. I’d like to identify the human transcriptome of this mix of reads by mapping it to the human genome. What tool can I use best and what (stringency) settings should I apply to get the most of human transcripts mapped?

2

Hi @Henk

The mapping tool included in most Transcriptomics tutorials, HISAT2, should work fine for this. But RNA Star can be adjusted to be the most sensitive as far as I know. You could even try both and compare.

For parameters, see tutorials then tune up the stringency as needed.

HISAT2 → https://training.galaxyproject.org/training-material/by-tool/iuc/hisat2.html

RNA Star → https://training.galaxyproject.org/training-material/by-tool/iuc/rna_star.html

Consider supplying and filtering hits with a known reference annotation with either since you aren’t interested in discovery. Or, try with and without and review the novels versus the other species to make sure this is doing what you want.

Don’t miss the spliced alignment parameters with HISAT2. Not everything is at default, and that is where reference annotation is also input.

Hope this helps!

3

Hi Jennifer,

Thanks for your suggestions. Can definitely try both and compare outcome. Curious to see what turns up.

image001.png

image001.jpg

1 Like
4

Hi Jenny,

Using HISAT2, I obtained a list of datasets containing the unmapped reads. Unfortunately, I exceeded my 250 GB and therefor couldn’t retain the datasets of mapped reads (which is not like a big problem as I’m only interested in the non-mapped reads, although I also would like to map the same samples with RNA STAR and still have a shortage in disk space). What I don’t understand is why I get 2x a list of datasets. One with (R) and one with (L) although I did the mapping on paired datasets. Can you explain this?

image002.png

5

Hi @Henk

For the space issues, this might help:

The UseGalaxy.org server has a brand new feature that will help: Scratch Space.

Go into your User > Preferences, and click into the Preferred Object Store for one way to set this up.

Scratch space grants 1 Tb of temporary short-term storage space to everyone, and you can move data between the original long term and this new short term space.

Brand new so no documentation yet – instead, just notice the little database icons and see the tool tips directly in the app. You’ll figure out how it works :slight_smile:

For the parts about the output, that is just how the underlying tool outputs the extra results. Remember that those won’t necessarily contain intact pairs anymore for certain combinations of parameters. Meaning, one end might map and the other not. If you are using paired, those would still be paired from what I know. If using collections, you could pair those again.

6

Thanks again for the tips and tricks. Unfortunately no ‘Preferred Object Store’ feature under ‘User>Preferences’ in the .eu server. I did notice the Scratch Space being available on the .org server when viewing my Storage Dashboard there, as well as the option ‘Preferred Storage Location’ under ‘User.Preferences’. However, not on the .eu.