这一新的突破性方法将使我们真正了解生物复杂性

Illumina公司的基本观点直接明确：生物十分复杂，为了揭示生物复杂性并借此对人类医学产生重要影响，需要大幅提升试验规模、大量削减每次试验的成本。某种程度上，这种理念成功了。

20年来，Illumina通过大力发展小型化多重分析，为基因组学革命注入了动力。无论是一开始的阵列，还是现在的新一代基因测序，都在纳米级别的实验中并列进行了数十亿次分析。这些工具及其应用为实现精准医疗带来希望，正在彻底改变整个医疗领域的实践。

2011年底推出了非侵入性产前检测，仅仅过了六年，每年检测数就高达200多万次。通过活组织检查或游离细胞分析进行肿瘤测序的方法正在改变癌症的诊断和治疗。改变已经足够明显，2018年首次批准了以肿瘤遗传特征（NTRK基因融合）而非传统的癌症起源界定（肺癌、结肠癌等）为基础的治疗型药物。

然而二十年后，精准医疗在很大程度上仍然只是希望。生物仍然非常复杂。似乎每当我们解开了其中的一部分复杂性时，就会发现更大范围的复杂性。

当我们开始了解某一特定基因的功能时，我们就需要了解该基因的网络，了解其RNA和蛋白质的翻译后修饰及其综合细胞调节机制，才有可能提出我们真正想问的问题：上述所有因素如何作用于疾病。

将整个人类基因组的测序成本从超过10亿美元降至1000美元以下，推动我们在这个还远没有答完的难题上取得了巨大进步。如果将成本降至100美元，会有更多发现；然而，这些发现将是增补性的而非革命性的。科学家在拓宽人类知识库的过程中，要求他们使用的工具采用了正交法，具有新功能。我们认为有两种方法符合该标准：分辨率的显著提升、曾经独立的学科和能力进行整合。

分辨率革命：单细胞基因组学

直到前不久，基因组学还一直依赖于大块组织样本的分析。虽然这种方式可以回答一些重要问题，但忽略了关键的生物信息。任何活组织检查都包含多种细胞类型。

人们已经知道（而且还在计数）有200余种人体细胞类型。把样本中所有细胞混在一起进行整体分析会导致不同细胞的功能被混为一谈，也会忽略不同细胞类型的比例差异。这种分析只能揭示最明显的信息。单细胞基因组学——能够在全基因组范围内单独分析每个细胞——为我们了解生物复杂性提供了有力的透镜。

10x Genomics（以及其他公司）正在开创单细胞基因组学的新方法，扩展可以在单细胞水平上进行的分析类型，包括DNA测序、RNA分析、表观遗传学和免疫学等。由于这类平台意义重大且发展迅猛，在整个研究生态中的应用也愈加广泛，今后必将开发出更多单细胞分析的应用方式。

单细胞基因组学在细胞层面的洞察力才开始真正展示出这种方法的力量；例如，新发现了一种罕见的呼吸道细胞类型（肺离子细胞），这种细胞被认为在囊性纤维化中发挥了重要作用。

即便如此，甚至还有其它方面的信息可以强化实验——这些细胞在组织中的位置和相互作用。这类空间信息对于理解某些疾病至关重要。

传统上，组织分析一直属于病理学范畴——用有一两个标记的薄片。未来工具开发的一个重要领域将是对具有完整细胞空间定位的组织进行基因组层面的分析，从而在组织层面上对细胞进行基因组分析。

整合的力量

尽管整合DNA、RNA和蛋白质数据的想法已经讨论了十多年，却囿于数据分辨率不足和统一分析技术的缺乏。但是最近的一些进展正在突破这些局限；现在可以对同一样本在单细胞层面上分析其RNA转录和表观遗传变化，从而深入了解表观遗传学如何影响转录。

同样，可以在单细胞层面上确定抗原以及抗原绑定的特定免疫受体的序列，为实现免疫建图、未来治疗和通用诊断测试创造了机会。

很快，整合将不再局限于这种“只读”性质的多层组学整合分析，而是会结合CRISPR（一种基因编辑技术）等强大的生物“写入”功能。通过将CRISPR与基因组工具以及单细胞分析相结合，科学家将能够并行不悖地进行写入（编辑DNA）、检测（分析某些生物学输出）、读取（测序），查询多项读数（DNA、RNA、蛋白质、表型）。

Perturb-SEQ是这种整合的早期例子，利用Perturb-SEQ进行多重分析时，会在单个细胞中扰动数万个单个基因，通过单细胞RNA分型来分析这种扰动产生的表型结果，从而实现了全面分析基因功能的功能性基因组学。

这种整合的下一步是在单个细胞中10000个不同位置上进行各不一样的基因写入，测试表型变化，其中包括单细胞RNA分型后基因表达的改变等。

迄今为止，生物写入一直费钱费力，仅限于大规模项目。然而，正在开发用于单细胞多重CRISPR编辑的桌面仪器，从而确保研究人员能够整合单细胞DNA读写。我们很快就能感受到在闭环系统中整合DNA读写的作用，还将看到这种做法会大大提升学习速度。

这些新方法将推动医学领域取得激动人心的重大进步，同时也突出强调了生物一直以来极度复杂的特性。毫无疑问，未来20年将出现的机遇和进步将比这20年更令人兴奋。（财富中文网）

约翰·施蒂尔普纳格尔是Illumina和Ariosa Diagnostics的联合创始人。此外，他目前担任10X Genomics董事会主席。布莱恩·罗伯茨是Venrock的合伙人。

译者：Agatha

Our founding thesis at Illumina was straightforward: biology is extremely complex and to unravel that biology, and thereby dramatically impact human medicine, would require a much larger scale of experimentation at an exponentially cheaper cost per experiment. And it worked, sort of.

For 20 years, Illumina has powered the genomics revolution through dramatic miniaturization and multiplexed assay development. First with arrays, and now with next-generation sequencing, experiments are conducted on the nanometer scale with billions of assays occurring in parallel. Those tools, and their applications, created the promise of personalized medicine and are revolutionizing entire areas of medical practice.

Non-invasive prenatal testing was introduced in late 2011, and only 6 years later more than 2 million tests are run annually. Sequencing of tumors, either via biopsy or through cell-free analysis, is changing cancer diagnosis and treatment. Enough so that 2018 saw the first initial drug approval for therapeutic usage based on tumor genetic signature (NTRK gene fusion) rather than the traditional delineation of cancer origin (lung, colon, etc).

Still however, two decades later, the promise of personalized medicine is primarily just that, a promise. Biology remains very complex. It seems that every time we unravel a portion of that complexity, we uncover more complexity.

As we start to understand a particular gene’s function, we then need to understand that gene’s networks, its post-translational modifications at the RNA and protein level, and its complex cellular regulation, before we can even get to the question we want to ask: how all of that impacts disease.

Reducing the sequencing cost for a whole human genome from more than $1 billion to under $1,000 has driven enormous progress on this very incomplete puzzle. Reducing it to $100 will generate additional discoveries; however, those discoveries will be incremental, not revolutionary. Scientists require orthogonal approaches and novel capabilities in the tools they use to catapult forward our knowledge base. We believe that two approaches fit this criteria: dramatic improvement in resolution, and the integration of previously disparate disciplines and capabilities.

A Resolution Revolution: Single Cell Genomics

Until very recently, genomics had relied on the analysis of bulk tissue samples. While important questions were answered, bulk analysis ignores critical biological information. Any biopsy is comprised of a variety of cell types.

More than 200 human cell types are known (and counting). Blending all of the cells of a sample together for bulk analysis obscures function at the cellular level and ignores proportional differences in cell types. With this type of analysis, we are only able to reveal the most obvious information. Single-cell genomics – the ability to analyze each cell individually on a genome-wide scale – is now providing the lens needed to match biological complexity.

10x Genomics (and others) are pioneering single-cell genomics approaches and expanding the types of analysis possible on the single cell level, including DNA sequencing, RNA profiling, epigenetic discovery, and immunology. More and more applications of single cell analysis will be developed now that this important platform has become robust and its usage is proliferating across the research ecosystem.

The cellular level insights from single-cell genomics are really just starting to demonstrate the power of this approach; for instance, the identification of a rare airway cell type (pulmonary ionocyte) now deemed to be important in cystic fibrosis.

That said, there is even another dimension of information to augment these experiments – how these cells are positioned and interact in the tissue. This spatial information will be vital to understanding some diseases.

Traditionally, the analysis of tissue has been the domain of pathology – thin slices stained with one or two markers. An important future area of tool development will bring genomic-level analysis to tissue with intact spatial positioning of cells, thereby allowing genomic assays to be run on cells at the tissue level.

The Power of Integration

While the idea of integrating DNA, RNA and protein data has been talked about for over a decade, this data has suffered from both a lack of resolution as well as unifying assay technology. However, recent advances are overcoming these limitations; it is now possible to analyze, at the single-cell level in the same sample, both RNA transcription and epigenetic changes, providing an insight into how epigenetics affects transcription.

Likewise, one can determine both the antigen and the sequence of the specific immune receptor to which the antigen binds with single-cell discrimination, opening up the opportunity for immune mapping, future therapeutics and universal diagnostic tests.

Soon, integration will expand beyond these multi-analyte biological “read only” assays, to incorporate powerful biological “writing” capabilities such as CRISPR. Integration of CRISPR with genomic tools and single-cell analysis will allow scientists to write (edit DNA), test (assay for some biological output), and read (sequence) in a parallel fashion, interrogating multiple readouts (DNA, RNA, protein, phenotype).

An early example of this integration is Perturb-SEQ, where, in a multiplexed assay, tens of thousands of individual genes are disrupted in single cells with the phenotypic results of that disruption being analyzed through single-cell RNA profiling, enabling comprehensive functional genomics.

The next step in this integration will be to write uniquely at 10,000s of different locations in single cells, then test for phenotypic change, including changes in gene expression through single-cell RNA profiling.

To date, biological writing has been an expensive and manually laborious process confined to large-scale efforts. However, desktop instruments are in development for single-cell, multiplexed CRISPR editing to enable researchers to integrate single-cell DNA reading and writing. We will soon appreciate the power of integrating DNA reading and writing in a closed loop system and the dramatically faster pace of learning that will result.

These novel approaches will drive crucial and exciting progress in medicine, while underscoring the continued enormity of the scale of biological complexity. The opportunities and advances of the next 20 years will undoubtedly be even more exciting than the last.

John Stuelpnagel is a co-founder of both Illumina and Ariosa Diagnostics. Additionally, he currently serves as Chairman of the Board of Directors of 10X Genomics. Bryan Roberts is a partner at Venrock.