ASHG Notebook: Bruker PaintScape Illustrates Chromosome Architecture

I spent last week at ASHG in beautiful Montreal.  I've scribbled many notes in my notebook, and am planning to emit a series of pieces each focused on a single topic.  Some of these may be brief, but I hope still interesting.  Far too many of these cover topics I had similar notes on from AGBT that never made it into pieces - such as the following on Bruker's PaintScape instrument.

Back at AGBT the Bruker PaintScape technology caught my eye, but it because one of numerous possible pieces which were iced by procrastination.  I caught up with science lead Jude Dunn at Bruker's ASHG booth and he kindly walked me back through the tech.  

PaintScape is an approach to visualizing the 3D structure of genomes in individual cells.  Inputs are fresh cells or tissues.  These are hybridized with cocktails of barcoded probes which are designed to generate signals at differing levels of resolution, with some lighting up broad regions and others specifically painting individual loci of interest.  Bruker has two probesets available or in active development, ChromoPanel with 400 targetws and an OncoPanel with a planned set of 1000 targets. Currently all probes are directed against human DNA; I suggested probesets against Epstein-Barr Virus might be interesting given the growing interest in EBV in chronic neurological disorders.

Because this jebFISH data is collected with X,Y and Z information, the distances between regions can be computed for each cell in the field-of-view.  The software generates correlation maps showing the distribution of distances between all the feature types.  This can then be used to detect a wide range of chromosomal abnormalities or aberrant folding. Extrachromosomal DNA can be resolved from chromosomal structures with ease.  Dunn showed me example images which showed the gene BRIP1 clearly existing on the chromosome but also as ecDNA with no other probes lighting up the ecDNA.

The technique that first pops into my mind for "nuclear three-dimensional architecture" is Hi-C.  Dunn pointed out several key differences between PaintScape data and that from Hi-C.  Because Hi-C relies on proximity ligation, it is biased towards close interactions.  To detect longer and longer range interactions requires deeper and deeper sequencing data until it isn't really practical.  PaintScape imaging inherently computes distances for all pairs of probes, and so freely generates both short and long range data.  

PaintScape is also computing on individual cells, whereas most HiC is a bulk method.  Yes, single-cell HiC exists, but the number of contacts mapped per cell is very low.  By imaging individual cells, PaintScape can dissect the heterogeneity of chromosome architecture across cells in a sample.

PaintScape is a high content, not high throughput, instrument.  Each slide requires 12 hours to process, though this depends on the number of fields of view imaged.  The system has a list price of $600K and consists of three benchtop units.  The fluidics and imaging units must be next to each other, but the sample prep instrument can be positioned anywhere.

So PaintScape is a method with high-resolution but low-throughput, requiring an expensive single use box.  But if you wish to reveal chromosome architecture in individual cells, PaintScape would seem to be in a gallery all its own.