OVERVIEW
On February 1, 2021, we followed-up with a molecular cell biologist that started working on single-cell genomics in 2010. We last touched base with this contact toward the end of August (2020). He characterizes his institution as a research community with ~30 faculty, but he also operates his own research lab. Federal and state funds support the main lab, which has a primary mandate on single cell research, but also the broader applications in single cell human biology. They use both 10x Genomics ($TXG) instruments and consumables often, as well as Illumina ($ILMN) sequencers. Some of his colleagues are working with PacBio ($PACB) or NanoString ($NSTG) equipment as well, but he is not. For research programs on human solid tumors, early human implantation, IPS, and embryonic stem cell differentiation schemes, among others, he prefers TXG.
Takeaways
-
The lab saw a very strong volume finish to 2020, both internally and among external customers.
-
Visium resolution needs to improve to single cell dimensions below their current 55 micrometer resolution.
-
Penetration can climb significantly higher, adoption is likely to parallel PCR or flow cytometry, both widely adopted.
CALL NOTES
We recall that your labs never shut down but there was a slowdown back in 2Q20... how is business today? How was the 4th quarter?
- That’s right, we handled volume for labs that had shut down (e.g., people would send samples for processing - frozen brain for single nucleus RNAseq) and were generating data the whole time. 4Q was one of the busiest quarters, ever. We’re just being cautious with [physical] distancing in the lab.
- We test everyone every week (that goes to the site) - we were in the early phase (front-line workers) so everyone is vaccinated (I get my second shot this week).
- We also handled - and continue to - a lot of clinical tests for the state and have samples coming through for research as well.
- It’s not really work being made up either. We’ve been on an increasing rate of business year-on-year for the last two years because of more single cell.
Looking at single cell and all the project terms in NIH grants, it looks like there’s rapid growth but it’s very underpenetrated. Where do you see it now?
- RFAs (requests for applications) for single cell capabilities are still growing. The scope is broadening too. I was discussing a study on single cell view of senescent cells in tissues and organs last year that was a pretty big grant. Are there drugs to treat those cells… anti-aging? Those cells haven’t been well-described in vivo, and researchers want to know the biomarkers, etc.
- There are multi-year programs in single cell like the Human Biomolecular Atlas Program - a single cell resolution atlas of humans. Paralleling the Human Cell Atlas.
- At our academic center, there’s growing interesting in single cell, so the user base is growing. It’s a long way from penetrated. Best guess - in molecular cell biology labs (specialized), 10-20%, but it’s more like single digits overall.
Do you have a view on NanoString’s (NSTG) GeoMx DSP?
- Yes, I hear from them often. I don’t view it as single cell resolution at all (100 micron). In that context, I haven’t pursued it. That said, we are running some spatial transcriptomics single cell using GeoMx with a partner to do a direct comp (presumably to TXG’s Chromium, but it was not specified) - a direct side-by-side comparison of advantages and disadvantages. There’s no data, yet.
- Beyond GeoMx, we know NSTG is developing the Whole Transcriptome Atlas and another system – the Spatial Molecular Imager (SMI) – that’ll be ready this year and in 2022, respectively. The latter will be higher resolution - truly down to the single cell, which looks pretty interesting, in my opinion.
- I think clinical practices are picking up the GeoMx because it’s more along the lines of a pathologist’s approach. However, to gain new insight into a tumor, you need single cell resolution. To research cell-to-cell communication, you need single cell resolution too. That's the drawback of the DSP system (again, the resolution).
- Will it be too late in a year?
- It’s hard to say. There are many spatial transcriptomics methods being developed, but a good number are rudimentary. If the SMI is competitive, it may not be too late.
What are your updated thoughts on 10x Genomics’ Visium?
- I haven’t seen improvements yet – I’m sure they must be working improving from their 55 micron resolution. There’s one new application combining it w/ antibodies, hashing it … looking at the protein and transcriptome at same time.
- A lot of people question the sensitivity, but Visium’s spatial transcriptomics is pretty impressive.
Is 10x the standard? Do you see anything making it less useful (in the context of them selling more consumables)?
- Yes, to the first part. No, to the second. 10x is still the go-to, it generates great data, even compared to spatial transcriptomics. We still get deeper data. It’s hard to beat TXG’s Chromium if you want that level and depth of profiling.
- Also, Chromium is such a good product – it’s always being improved upon. Our reagent spend is going up ~25% y/y (we recently placed a purchase order for $1.5M), the majority of which (~80%) is for our use (vs. work for others).
- There’s a lot going on – imaging mass cytometry, conjugated antibody multiplex method plus Visium spatial and single cell ATAC, etc. – we’re running everything and will compare to see what yields the best results. It’s very busy.
- TXG’s new reagents at a lower cost will be great – people are doing multiplexing already. It opens up to labs just doing cell culture experiments where throwing $3k at a sample isn’t feasible. For example, you could apply this routinely to regenerative medicine – pooling 10 samples on a single channel (Chromium) wouldn’t have been done. The amount of insight we’d get – differentiating system is profound. This will be a useful, heavily adopted tool.
- It’s single transcriptome cytology vs. FLOW cytometry using the 10x system. BD does FLOW cytometry on 30 antibodies - closing in on 35-40. It’s only 10s of markers vs. thousands of genes and how they are expressed.
- Once the Atlas is done, we will reach a state where you can target single cell transcriptomics – i.e., “what are the best 1k genes to define a blood cell?” Use Chromium, targeted approach, and drop the amount of sequencing. There’s a Lupus blood study, 400k transcriptomes - good view of transcriptome dynamics cellular heterogeneity - conducting targeted panel is cost effective and we can look at the immune signatures of disease state, autoimmune reaction in tissue, revealed in blood.
What are your current views on short vs. long reads (Illumina (ILMN) vs. PacBio (PACB))?
- I have some colleagues that are very + on PACB and like the Sequel II/IIe (they are big users), but I’m not a PACB user. For the depth of sequencing and price nothing rivals ILMN.
- Right now, nothing beats being able to sequence cheaply and deeply. After the recent price drop – yes, they control the market – nothing is comparable. The cost is lower and throughput greater.
- Illumina allows sequencing costs to go down. It was a significant price drop and density increased, so we get more sequence out of the platform. ILMN dominates and can decide the price.
- That said, long reads are great. There’s Nanopore too, but throughput is the drawback, especially relative to ILMN.
Has anything about Oxford Nanopore changed in the last 6-12 months?
- I’m not an Oxford Nanopore user either but the PromethION system has improved throughput. It’s still not as competitive as Illumina’s systems, in my opinion. That said, it depends what a researcher is doing or where they are going after -> assembling new genomes, combinations of short reads plus long read… that can be advantageous.
Last we spoke, you mentioned neurology as an area of new demand (for single cell) – anything new?
- Anything to do with the brain is a big area right now. Cancer is another big one.
- The large NIH funded programs to build atlases – HTAN (Human Tumor Atlas Network) & Human Biomolecular Atlas Program (HuBMAP) – and mice modeling are big drivers of single cell demand.
- If you profile the primary tumor with transcriptomics you can build models and improve a mouse’s ability to host a tumor or add to organoid culture systems.
- On the back end, there’s exponentially more data. TXG provides user-friendly software, but it’s not appropriate for large studies. The HuBMAP consortium = large computational and database efforts. Data portability across platforms is important.
Do you have a view on cfDNA and Grail?
- When you have a patient’s tumor, you know what you’re screening for (MRD), whereas screening is harder. Grail’s 99%+ specificity is good but not sure where it’s coming from. Capturing circulating tumor cells (CTCs) are more valuable to generate models. However, it’s challenging to capture CTCs unless it’s small cell lung cancer (most abundant, likely metastatic).
- Back to single cell – being able to profile a primary tumor with single cell – seeing the cancer cell and transcriptome, other cells communicating with it to repress or kill it – is exciting.
- Can single cell help accelerate things? How close are we?
- There have been studies – e.g., w/ bladder cancer – where we see somatic mutations in “normal tissue” - precancerous deemed normal tissue. If you sample multiple regions, you can find different subsets of mutations. Some enriched in bladder cancer genomes, but many are not found. Is it a first step toward a cancerous state that requires further mutation to progress? It’s surprising – you’d think normal tissue is normal.
Please reach out to with any feedback or inquiries, questions for future field checks, or requests for underlying data.
Thomas Tobin
Managing Director
Twitter
LinkedIn
