Recently, a collaboration bringing together two of the hottest technologies in life sciences: single-cell transcriptomics and pooled CRISPR screening, was announced. This development allows the simultaneous measurement of gene expression changes in single cells on the 10x Genomics’ platform using reagents and workflows from MilliporeSigma. Innovations like these have continued to push the envelope for next-generation sequencing (NGS) in terms of both its technical and scientific capabilities.
NGS is a method to sequence DNA, RNA, and epigenomic features that has expanded on the first-generation Sanger method of sequencing. Unlike Sanger, NGS methods are massively parallel, high-throughput, and performed at high speeds and lower costs, and those limits continue to be pushed with each new generation of instrumentation launched. NGS technologies can be used for various types of sequencing such as whole-genome and whole-exome DNA sequencing, methylation sequencing, whole-transcriptome sequencing or RNA sequencing (RNA-seq), single-cell sequencing (scSeq), targeted gene sequencing, and more. Whole-genome sequencing is typically used to identify pathogenic variants underlying human diseases like cancer, which are associated with changes in the coding and non-coding regions of the genome. In contrast, exome sequencing is used for Mendelian disorders or rare diseases that are often involved with changes in specific coding regions of the genome.
Factors underlying buying decisions: What’s in the box?
Sequencing has become fairly routine and user-friendly, but it is clear that no one technology can answer all the biological questions. Hence, innovative new products are regularly launched and continuous updates are made over the lifetime of those products to improve speed and sensitivity. When and what to invest in, therefore, is an important decision. “Of course, the cost will always be an important factor but what you are looking to study (the application) and how many samples you are likely to analyze (the throughput) will more or less determine what technology you should invest in,” says Shrikant Mane, Ph.D., Professor of Genetics at Yale University and Director of the Yale Center for Genome Analysis.
Each type of sequencing technology has its own unique advantages, and vendors offer a range of products that cater to different budgets, throughput, and application needs. NGS platforms are also continually updated to improve automation, ease-of-use, analysis time, and to offer simultaneous measurements of different parameters. Some sequencers are low throughput and cost less, which is ideal for labs that process only few samples. Similarly, there are state-of-the-art sequencers that are expensive and process a large number of samples to keep things cost efficient. Some sequencers perform short-read sequencing using 300–400 base pair fragments of DNA or RNA, which is good for looking at gene-expression changes. However, for studying alternate splicing or structural variations you need to sequence around 1000 base pairs, and those instruments tend to be more expensive.
Image: The PacBio Sequel II System provides scientists with access to high-throughput, cost-effective, accurate long-read sequencing.
“Long-read technology is important for a comprehensive understanding of the genome as it reveals more genomic context through the ability to sequence high and low GC areas, offer important phasing information, and provide full-length RNA transcripts,” says Marty Badgett, Senior Director Product Management at Pacific Biosciences. “These advances are already showing promise in areas like metagenomics where scientists really need long-read sequencing results to provide a better understanding of the population being studied.”
KNOW WHAT YOU WANT
Like with all other purchases, the first step to evaluating and investing in NGS starts with empowering yourself with knowledge—about the products and how they will be used.
- There are several resources, easily accessible on the Internet, that give you relevant information and practical know-how in a manner that is easy to digest
- Familiarize yourself with the technology and the workflow, understand what is needed to complete a successful experiment
- Make a list of what is available commercially and what you may have to put together yourself
- Find people who have been doing sequencing for years. There is no substitute for knowledge gained from years of experience. Take the time to listen and learn about what worked for them and what didn’t
- Join NGS-specific user groups and discussion forums to hear about the latest developments in the field and persisting challenges
GET WHAT YOU NEED
- Based on your research, make a list of vendors you would like to talk to
- Be prepared with questions to ask the vendor and have samples ready to analyze if they are offering a product demonstration
- Test-drive a few things related to the sample prep, instrument and reagent use, and software manipulations to get a feel for how user-friendly the platform is
- Take the time to get details about all the hidden costs. The cost of analysis per sample should include everything from sample prep to service contracts for routine maintenance, not just the cost of the instrument. It may help to talk to people at core sequencing labs who routinely calculate cost per sample for their clients
- Knowing what new technologies or upgrades are coming up may help in budgeting for a new instrument, software, accessories or add-ons
- Before signing on the dotted line, think about where the instrument will be placed in the lab, who will use it, how often will it be used, and how you can maximize the return on investment
Factors underlying buying decisions: Thinking outside the box
“Being a core lab, the cost of sequencing per sample is very important to us,” says Mane. “We avoid repetitions and mishaps by ensuring that the quality of the sample is good, the machines are well maintained, the protocols are rigidly followed, and the technicians are well trained.” The source of material to be analyzed (cells versus fresh or formalin-fixed tissue) and the amount of material in the sample available for sequencing does not play much of a role in determining which technology to invest in, but it does impact the options for sample-preparation reagents and kits. “We choose our sample- preparation kits based on reliability and costs. Sample processing and quality control is critical or else you will not get good results even after investing in the right technology.”
“Sample preparation is all about the little things done right,” says Luciano Martelotto, Ph.D., Scientific Director of the Single Cell Core at Harvard University. “It’s the most difficult and the most under-appreciated aspect of single-cell genomics.” The type of sample, its origin, its composition, the fraction of the material of interest (rarity), and time taken to collect and process the sample can all affect the sample quality. “It’s important to have the lowest number of steps in sample preparation to mitigate potential artifacts or biases,” says Martelotto. “In scSeq it’s not the sequencing but the sample prep and the encapsulation of cells that is the most expensive part of the experiment.” Hence, finding the right reagents and kits and paying attention to the details of the protocol is very important.
After sequencing comes data analysis, which has its own set of challenges. NGS data analysis requires scalable computational workflows for data processing that involves quality checks and assigning the raw sequence reads to a reference sequence (this can be a genome, transcriptome, or specific sequences of interest). This is followed by application-specific downstream analyses like variant calling for WGS studies or clustering and visualization for scRNAseq to interactively query the data and see which aspects are most relevant to the biological question being asked.
In the case of single-cell transcriptomics, thousands of genes are profiled in individual cells, with thousands of cells within a sample. Each experiment may also have multiple samples collected under different conditions or over a period of time. This generates large amounts of data. “The number of samples, cells, and reads in an experiment determines the amount of data to be analyzed. In order to be efficient, the analysis needs to be done in a high-performance computing (HPC) environment,” says Shannan Ho Sui, Ph.D., Senior Research Scientist in the Department of Biostatistics at the Harvard T.H. Chan School of Public Health and Director of the Harvard Chan Bioinformatics Core. In addition, NGS data are high dimensional, which means that the number of features often exceeds the number of observations, leading to statistical challenges.
Factors underlying buying decisions: Buy versus DIY
Single-cell sequencing experiments are usually complex and expensive endeavors. Hence, making the right experimental decisions can save both time and money. “Nowadays, you can buy kits and instruments to aid your single-cell experiments; everything comes perfectly packed in a box, ready-to-use,” says Martelotto. “However, if a commercially available platform cannot fully address your biology questions, then you may have to do some modifications or tweaks to suit your needs, and even develop your own reagents and methodologies.”
Most vendors, while generous in discussing and sharing their knowledge and expertise, will not support changes made to their platform, which could lead to problems in the functioning of the equipment. Similarly, home-brewed reagents and protocols require optimization, whereas commercially available reagents and kits come standardized and are fairly robust.
The choice to do-it-yourself also depends on the application. “If there is a new protocol that works better, it can be tested and implemented in a research setting. In a clinical setting everything has to be evaluated, approved, and regulated before changes can happen," says Martelotto.
Ho Sui feels that in order to be at the bleeding edge of data analysis, her core lab needs to use tools as they are being developed in academia. “The technologies and methods are rapidly changing so we are always evaluating new tools that approach biological questions in interesting and statistically robust ways. Often, as these approaches mature and emerge as best practice, they are integrated into commercial software; hence, in some ways, using commercially available software puts us behind the curve compared to using tools that are in development.”
However, the benefit of having commercial options is that the methods are validated, tuned, and standardized, which is important for smaller labs and companies that may want something that is easy to use and systematic. “We find that many of our collaborators want to be most engaged in the data visualization aspect of the data analysis process, after much of the processing and statistical analyses have been performed, and I think that’s where commercially available tools can add a lot of value,” says Ho Sui.
Ultimately the choice of whether to buy or design comes down to the sample, application, budget, skillset available, the type of instrument being used, the cost of sequencing, and the experience of the user. “In some situations you can collaborate with the vendor to make changes to the platform or protocols, but that change has to be something that will appeal to a wider customer base and be financially viable for them,” says Martelotto.