Entering edit mode
10.1 years ago
spiralguru
•
0
I created single-cell RNAseq libraries from neurons and sequenced them at depths of ~10 million reads per cell. Almost 90% of my sequences have no hits according to fastq screen, and only 2-10% map to the mouse genome (data for one sample attached). As someone who just started doing RNAseq analysis, I cannot tell whether this reflects a shortcoming of the kit I am using or that I have sequenced the libraries too deep and I am getting unexpectedly high noise. Any thoughts and suggestions?
are your samples barcoded? have you removed them
have you trimmed the adapters?
before mapping run fastqc to check how does your sequences look like
Actually, the fastq screen run (data shown above) was done before trimming the adapters (fastqc was done and that showed quality scores between 30-40 for all 50 bases), but I've trimmed the adapters off and tried aligning using tophat2. I get alignment to the mouse genome (mm10) ranging from ~2% to ~15%. I had one sample with 10 cells pooled, and that showed 40% alignment. These numbers, especially the single cell reads sound, awfully low.
Try blasting a few reads and see what you get.
There are definitely sequences that map to mouse genes. Cufflinks detected 1000-8000 genes per single-cell library with FPKM >1. But over 90% of my reads don't map anywhere based on both fastq screen and tophat. Am I just being paranoid? Or did I create this situation for myself by sequencing too deep? (I've read recently that you tend to get amplification of 'unreal' things when you sequence too deep if your library complexity is relatively low, but haven't seen others report 90% no hit).
Perhaps you got some junk, but you'll only know by blasting a few of the unaligned reads. If they don't align to anything, they're probably just spurious amplification products.
What is up with the E. coli reads? Did you do any QC on your RNA before sequencing?
Yeah, looks like microbial contamination is an issue with my samples. I didn't run any QC before sequencing unfortunately. How does one do that given the tiny amount of RNA one gets out of each cell? Any advice would be helpful.