Chromium 5’ Next GEM single cell STARsolo

Question

[hood821]

Hello!

I've run STARsolo in practice with the neurons_900 (mouse) dataset from 10x, and I am able to get the correct amount of cells. Then I tried running on my real samples (Mouse), which are expected 5000 cells, I get ~148 cells, the assay is: Chromium 5’ Next GEM single cell kit from 10X with the targeting of 5000cells/sample (100cycle kit with an S1 flowcell (Novaseq) as suggested by 10X).

For STARsolo:

I built the genome index, using: Mus_musculus.GRCm39.112.gt & Mus.musculus.GRCm39.dna_sm.primary_assembly.fa I used whitelist: 737K-august-2016.txt

STAR --genomeDir ~/10x_Test_Data/ref --readFilesIn ~/10x_Test_Data/neurons_900_fastqs/neurons_900_S1_L001_R2_001.fastq ~/10x_Test_Data/neurons_900_fastqs/neurons_900_S1_L001_R1_001.fastq --soloType CB_UMI_Simple --soloCBwhitelist ~/10x_Test_Data/whitelist/737K-august-2016.txt --outFileNamePrefix ~/10x_Test_Data/output/ --outSAMprimaryFlag AllBestScore

This worked, but when I ran my real samples, I couldn't figure out which whitelist to use with this Chromium 5' Next GEM kit. So I tried different whitelists and tried no whitelist and I always get ~100-200 cells instead of the 5000. I don't know what I am doing wrong. I think it is the assay, so I was going to try using cellranger, but i would like to get StarSolo to work.

Answer 1

[dsull]

Take a look here:

STARsolo ; Reads are NOT mapped to transcriptome

Follow the code in the comment that was written in response to my answer.

[hood821]

I tried running your code, and it didn't work sadly. I did run it with cellranger and i am getting 6008 cells, and the cell ranger output said in the autodetected chemistry : Single Cell 5' R2-only. I would like to get STAR to work, any thoughts on how to adjust the input?

[dsull]

Saying “it didn’t work” gives me no information… I can’t help you if all you say is it “didn’t work”. What’s the error message?

Answer 2

[swbarnes2]

You might not be doing anything wrong. Your samples might not have as many cells as you wanted. Try using something like umi_tools to empirically work out how many cell barcodes are present.

[dsull]

I wonder if the neurons_900 samples are using the 5' chemistry. When I used it in the past, it was 3' chemistry -- which is processed differently.

[hood821]

I did run it with cellranger and i am getting 6008 cells, and the cell ranger output said in the autodetected chemistry : Single Cell 5' R2-only. I would like to get STAR to work, any thoughts on how to adjust the input?

[swbarnes2]

If you poke around the celleanger folders, you should be able to find the whitelists of cell barcodes. Compare those to what you've tried.