Note: This pages provides a longer overview of the 10X Genomics On-Chip Multiplexing assays for those who are interested in more detail on how this assay works in comparison to the standard 3′ and 5′ Gene Expression assays. For the condensed version focusing more specifically on the strengths and limitations of the OCM approach, please see this page.
10X Genomics recently launched new versions of their flagship 3′ and 5′ Gene Expression assays integrating an approach they call “On-Chip Multiplexing” or OCM. The OCM assays feature a relatively very low per sample reagent cost compared to other 10X products, and these versions of the assays have some distinct advantages over other multiplexing options, but they also have their own unique limitations. The purpose of this post is to give an overview of this approach to clarify the extent of both the advantages and limitations. Some general background on how 10X Genomics assays function is provided below in order to explain what makes the OCM assays unique; the general strengths and limitations of OCM is expanded on in much more detail at the end of this post.
10X Genomics assays isolate single cells by microfluidics
The 10X Genomics process for partitioning cells involves three key components: the cells; gel beads, which are covered in a lawn of barcoded primers that will be used to synthesize cDNA from the captured mRNA; and partitioning oil, which is used to generate the aqueous cell-gel bead droplet units (referred to as “gel beads in emulsion”, or GEMs). The actual capture process occurs through microfluidics – separately, the cells (pre-mixed with reverse transcription master mix), the gel beads, and the partitioning oil are loaded into a single use plastic chip provided by 10X Genomics. The 10X Genomics instruments required for this process effectively acts as a fancy pressure pump, drawing these components through microfluidic channels on the chip to combine in a separate collection well. Once the GEM units have formed, the cells lyse and the gel beads dissolve, freeing the barcoded primers and the mRNA in order to initiate the reverse transcription reaction.
The chips that these materials are loaded into contain eight separated vertical lanes in the standard version of the assay. Up to eight samples are loaded into the chip, and each one of those samples produces an entirely separate GEM reaction pool that will go on to become its own RNA-seq library.
The OCM assays take advantage of existing barcode diversity and enables multiplexing by altering the structure of the microfluidic chips
The primers carried by the gel beads in these GEM reactions carry two unique barcode sequences. The first is a 16 bp sequence referred to as the “10X barcode” or the cell barcode. The second is a 12 bp sequence referred to as the unique molecular identifier, or UMI.
For all of the primers on a given gel bead, the cell barcode sequence will be identical, while all of the UMI sequences will be unique – so every reverse transcribed cDNA molecule will have a common cell barcode identifying the cell of origin and a unique UMI sequence that allows us to count the number of total transcripts for each cell. The cell barcodes being 16 bp in length allows for billions of possible barcodes; in practice, 10X Genomics uses known subsets of several million of these barcodes in each assay type, which their analysis pipeline will automatically recognize and pick out of the sequencing data. In the non-OCM versions of the assay, it generally does not matter if two samples in the same experiment have any overlapping cell barcodes in use among the cells captured from each sample. During construction of the sequencing library, each library will receive a third unique barcode known as the “sample index”, which is used to enable the pooling of libraries for sequencing. In order for two cells from different samples using the same cell barcode to be misidentified as a single cell, the two libraries would also need to be prepared with the same sample index AND be included in the same sequencing pool, which is extremely unlikely without serious human error.
The OCM assays make two notable changes to this process. The first is a change in the architecture of the chip. Instead of having eight completely separate lanes feeding into eight individual GEM collection wells, the chips used in the OCM assay group the lanes into sets of four. Each set of four lanes still contains separate wells for loading the cells, gel beads, and oils, but these pathways all feed into one GEM collection well, combining up to four samples into one GEM reaction.
The second difference is that each unit of the gel beads included in the OCM version of the reagent kit contains a pre-determined set of gel beads with a unique, non-overlapping list of cell barcodes. This is the key to the assay – when the samples are loaded into the chip, we know exactly what cell barcodes will be used for sample 1, what cell barcodes will be used for sample 2, and so on. This is why we are able to pool all four samples into one GEM collection well without doing any additional labeling steps in order to identify which original sample each cell came from in the eventual sequencing data.
Strengths and limitations of the OCM assay
The most substantial benefit to using the OCM assay is the price of the reagents. As of February 2025, the list price of a 16-reaction/sample kit for the standard, non-OCM 3′ Gene Expression assay is approximately $23,700. A 16-sample kit for the OCM version of the assay has a list price of approximately $9,000. Outside of this cost reduction in the main library preparation kit, additional “per library” consumable costs, including the aforementioned sample indexes and materials needed to QC the library at the cDNA stage and the final library stage, will also be lower since we are working with 1/4th as many libraries.
The other significant strength of this assay as a multiplexing option is that it avoids some of the more challenging aspects of using a cell labeling-based approach like antibody hashing. Namely, you do not need the relatively high numbers of starting cells (1M or more) that BioLegend recommends as input for their staining protocols, and since the cells do not need to be put through the additional extensive hands-on processing required to label them, we can get them loaded into the assay in a relatively fresh state and minimize the sort of cell health impacts that can potentially arise as the time between the initial collection of the cells and their capture in the 10X assay increases.
On the other hand, the OCM has a few notable limitations. These will be discussed in more detail below, but in short:
- The number of cells per sample we are able to capture is much lower.
- There is somewhat reduced flexibility in terms of the number of samples that can (or should) be run at once, especially compared to projects using the standard 3′ assay in our core.
- The OCM assay is incompatible with “sort & load” experiments where a very rare cell population (~10,000 cells or fewer) is sorted into a small volume of buffer and loaded directly into the 10X assay.
The most significant limitation of the OCM assay is a much lower cap on the number of cells that can be captured per sample. In the standard/non-OCM version of the assay, each of the eight lanes on the chip has a recommended maximum cap of 20,000 captured cells per sample. In the OCM assay, the maximum recommended capture target is only 5,000 cells per sample (meaning that the total combined GEM reaction is still capped at 20,000 captured cells). 10X Genomics sets this cap based on the “multiplet rate”, which is the proportion of GEM units that are expected to contain more than one cell. This multiplet rate is much higher in the OCM assay – putting in enough cells to target 5,000 cells per sample in this assay will result in the same ~8% multiplet rate that is seen when targeting 20,000 cells in the standard version. We can get around this limitation to some extent by loading samples multiple times – e.g. instead of loading four samples and getting 5,000 cells for each sample, we load two samples in two lanes each in order to capture 10,000 total cells from that sample – but this rapidly erodes the per sample cost savings for the assay and may still be undesirable if you are trying to capture a relatively rare cell population that would significantly benefit from being able to get a full 20,000 cells per sample.
In terms of the flexibility for running different sample numbers – for the 3′ Gene Expression assay, the Gene Expression Center stocks the 16 reaction kits. This allows our client base to run however many samples they wish to run for their experiment without needing to commit to larger reagent costs. If someone only wants to run one or two samples, they will only pay for one or two units of the library preparation reagents instead of needing to buy (at minimum) a four-sample kit. For the OCM kits, unless they become popular enough that we can be sure we won’t have reagents expiring unused, we would need to order a 16-sample kit specifically for your project and bill you for the full cost of that kit, even if you aren’t necessarily interested in running 16 samples (or even eight samples with double-loading). Additionally, the nature of the multiplexing in the kit is essentially “use it or lose it” since gel beads must still be used in a lane even if that lane receives only a mock sample (PBS) in the sample well. If you want to run 16 total samples, but are limited to only being able to prepare two samples at once (e.g. due to a lengthy sample prep or required sort), we would need to order multiple kits.
Lastly, regarding projects involving the sorting of a very rare cell population. This approach is possible in the standard version of the 3′ assay because the maximum volume of sample that can be put in to each lane is around 37 uL. This volume is usually pretty close to the volume of sample that we will have when roughly 10,000 cells are sorted directly into a strip tube with a relatively small nozzle (to minimize the buffer volume carried through per cell), including a 10 uL “buffer cushion” placed into the tube at the start to catch the sorted cells. In the OCM assay, since we are essentially breaking each GEM reaction into four separate inputs, the maximum volume we can load per sample is only 9.6 uL. This volume is far too small to be able to run these sorts of samples without trying a centrifugation step, which itself is likely to be extremely risky with so few cells.