Single cell library preparation, and when low input RNA seq is the better choice

Choosing library preparation kits, in the order the decisions come

1. Decide many cells shallowly or few cells deeply. Droplet chemistries profile thousands to tens of thousands of cells at a transcript end, which suits composition and cell type discovery. Plate based methods profile hundreds of cells with full length coverage, which suits isoform work, low expressed genes and anything needing the whole transcript.
2. Get the suspension right, because it is the whole experiment. A single cell suspension with high viability, no clumps and no debris is the requirement. Viability below roughly nine in ten introduces ambient RNA that contaminates every droplet, and dissociation stress itself changes the transcriptome, which is a well documented artefact.
3. Count and load deliberately. Loading concentration sets the balance between cells captured and doublets formed. Count accurately on the suspension you will actually load, not on an earlier aliquot, and accept fewer cells rather than a doublet rate that confounds every cluster.
4. Treat low input bulk as a different problem. Low input RNA sequencing from a few cells or a biopsy shares the amplification sensitivity but keeps bulk statistics. Where the question is about average expression in a small sample rather than about heterogeneity, low input bulk is cheaper, simpler and better powered.
5. Plan the analysis before the run. Ambient RNA correction, doublet detection, the integration method across samples and the clustering resolution all change the answer. Choosing them after seeing the clusters is how single cell studies produce cell types that do not replicate.

Dissociation and single cell dispensing are experimental treatments

Enzymatic dissociation at thirty seven degrees induces stress response genes within minutes, and those genes then appear as a cluster. Cold active protease methods and transcription inhibitors reduce it, and reporting the dissociation method is not optional.

Where tissue is difficult, single nucleus sequencing avoids dissociation stress entirely and loses cytoplasmic transcripts. For frozen archival tissue it is frequently the only option.

Replication in cDNA library preparation, the commonest design failure

Thousands of cells from one animal is one biological replicate, not thousands. Statistical claims about groups need several individuals per group, and a great many published single cell comparisons are pseudo-replicated.

Multiplexing samples into one run with hashing or genetic demultiplexing gives replication and removes batch effects at the same time, and is worth designing in from the start.

What to agree with a single cell RNA seq service

Who prepares the suspension, what viability they will accept, what happens if a sample fails quality control, the target cell recovery and depth per cell, and the files delivered. Suspension preparation is where most failures happen and it is often nobody's stated responsibility.

Agree that failed samples are reported before sequencing rather than sequenced anyway. Paying to sequence a library made from a poor suspension helps nobody.

single cell dna sequencing, and what it demands

Sequencing DNA from one cell means amplifying picograms, and every method's artefacts come from that: allelic dropout, uneven coverage and chimeras. Whole genome amplification chemistries trade uniformity against fidelity, and the choice follows whether copy number or point mutations are the readout. Cell isolation matters as much as the chemistry, since a doublet is two genomes and a dead cell is fragmented DNA. Analysis has to model the amplification rather than treat the reads as a bulk sample.

Common questions

Droplet or plate based sc RNA seq library preparation?

Droplet for thousands of cells and questions about composition and cell type. Plate based for full length coverage, isoform level questions and small numbers of precious or sorted cells. They answer different questions at very different costs per cell.

What viability does a single cell sequencing service need?

As high as you can achieve, and above roughly nine in ten for droplet chemistries. Dead cells release RNA into the suspension, and that ambient RNA is then captured in every droplet and has to be corrected computationally.

When is low input or FFPE RNA seq the better choice?

When the question is about average expression in a small sample rather than about heterogeneity between cells. It is cheaper, has better established statistics and does not pay the cost of single cell sparsity.

How many cells do I need?

Enough to see the rarest population you care about several times over, which follows from the expected frequency rather than from a round number. And enough biological replicates, which is the constraint single cell studies most often ignore.

Browse by service class

• Compare Gene synthesis on advertised prices