Entering edit mode
2.0 years ago
dew
▴
10
Dear guys,
May I know the typical mapping rate of the ncRNA mapping rate to the hg38.nc.fasta?
Thank you very much!
Best,
What exactly is hg38.nc.fasta and what kind of RNA-seq (total, polyA+ or Ribo-) are you talking about ?
Thanks, it's total RNA for RNAseq, the hg38.nc.fasta is Homo_sapiens.GRCh38.ncrna.fa.gz from ftp://ftp.ensembl.org/pub/release-98/ ; idea is to do the non coding RNA expression quantification after mapping. Thank you very much!
You should always map to a full and complete reference rather than a selected one as a selected one can lead to false-positive alignments. The aligner, in the absence of the true origin, will try to find the second best (but false) hit. Use the full transcriptome file, and then later subset to ncRNAs using a GTF as reference that contains these. The question you have cannot be answered, it really depends on the sample and library prep. I sometimes have RNA-seq with 95% mapping rates and that decreases the lower the input. In scRNA-seq you sometimes have 50% mapping while data are prefectly fine.
Thanks a lot for your guidance!
May I know if you have some experience with Kallisto, that can first map the reads to the whole genome, and then use the Kallisto quant -g option to designate the ncRNA GTF file to only quantifiability the ncRNA? Also, It seems the database/papers are more focused on the lncRNA instead of the whole ncRNA, so is it still necessary stick to the whole ncRNA or also focus on the lncRNA?
Thank you very much! Appreciate!
Is you want to use pseudo- or selective aligners such as kallisto and salmon it also makes sense to use the full reference transcriptome fasta file, this should include all annotated transcripts. I would always map to full, and then later once you have the counts subset to the genes or "RNA species" you want.
I second what ATpoint wrote. You should map on a complete reference.
That being said, for total RNA without ribodepletion and out of all reads mapped, you can expect a very large proportion of reads mapping to rRNA ( > 80 %), a few percents mapping to sno/sn/tRNAs, and in the order of 1% on lncRNA.
Thanks a lot! Appreciate all your kind guidance! All the best!