all information here is public
- Illumina Kit available:
-
Profiling vs analysis
- profiling focuses on counts against a reference, usually single end lib will suffice
- analysis aim to find sequence (e.g. novel fusion gene structure discovery), usually needs 2X reads than profiling (5-25M for polyA on human-sized transcriptome).
-
PE vs SE: read 1 better cov at 3'; read 2 better cov at 5', PE will decrease %dup
-
RNA input
-
RIN (for mammalian species only; >8 means good). Poly-A enrichment methods needs RIN>8 otherwise would have 3' bias, Ribo-depletion methods are more robust to low RIN, nbut needs 2X reads to achieve same coverage
-
Bioanalyzer/Fragment trace (look for size and unexpected peaks)
- all Illumina RNA kits use chemical fragmentation (divalent cations + heat)
- perfect prep BA trace
-
- perfect prep FA trace
-
- Peak max will be much larger than expected size, for PCR-free ligation (which results in forked adaptors given no PCR linearization)
- perfect prep RNA FA trace (peak will be very broad given no insert size selection), newer tech (Illumina RNA XXX) will have peak ~300-400 for polyA, 375-475 for Ribo0+
-
- Common problems:
- Overloaded chip
- No library peak
- large peak: insufficient bead clean up
- large peak: PCR artifact
- Adaptor dimer ~120bp, should be removed before putting on flowcell because shorter fragments cluster better so take up more resources, esp using patterned flowcell
- Overloaded chip
-
Quantify input: fluorometric method (e.g. Ribo-green) preferred, DNA might impact esp in FFPE, qPCR is most accurate
-
- Traces vs quant for different RNA kits
-
-
-
-
Stranded vs non-directional workflows
- Truseq RNA v2
-
- Truseq Stranded mRNA (dUTP in 2nd strand syn, pol will not pass U so only top strand is amplified)
-
- RNA index anchors, T-overhang results in low base diversity at first cycles
- Truseq Stranded total RNA
-
-
- Great notes on strandness of RNA-seq
-
Misc debugging tips