Erratum to: The real cost of sequencing: scaling computation to keep pace with data generation

Paul Muir, Shantao Li, Shaoke Lou, Daifeng Wang, Daniel Spakowicz, Leonidas Salichos, Jing Zhang, George M. Weinstock, Farren J. Isaacs, Joel Rozowsky, Mark GersteinGenome 2016 · DOI · OpenAlex

Abstract

This document serves as a formal erratum for the 2016 research article titled "The real cost of sequencing: scaling computation to keep pace with data generation." The authors identified a significant labeling error on the y-axis of Figure 4 in the original publication. In the initial version, the y-axis displayed the value "20" at every tick mark, obscuring the actual scale of the data. This erratum provides the corrected figure and re-evaluates the hiring trends it represents, ensuring the scientific record accurately reflects the growth of the computational biology workforce.

Introduction: The Computational Bottleneck

The original research addressed a critical challenge in modern genomics: the "computational bottleneck." While the cost of DNA sequencing has plummeted—dropping at a rate faster than Moore’s Law—the cost and complexity of the computation required to analyze this deluge of data have not followed a similar trajectory. This disparity creates a significant infrastructure and economic burden on research institutions.

To understand how the academic community has responded to this challenge, the authors analyzed faculty hiring trends. The hypothesis was that the explosion of sequencing data would necessitate a corresponding increase in human capital—specifically, experts capable of developing the algorithms and software needed for large-scale genomic analysis. Figure 4 was intended to visualize this trend, but the y-axis error rendered the magnitude of the hiring surge uninterpretable.

Methods and Data Sources

The data presented in the corrected figure is derived from a longitudinal tracking of faculty hires at 51 top-tier US universities.

  • Data Source: The dataset was sourced from jeffhuang.com/computer_science_professors.html, which tracks the career trajectories of computer science faculty.
  • Temporal Binning: To smooth out annual fluctuations and highlight long-term institutional shifts, the hiring data was grouped into three-year bins spanning from 1964 to 2014.
  • Correction Details: The y-axis has been updated to a functional linear scale ranging from 0 to 20, correctly labeling the "Number of faculty position hires."

Key Results and Analysis

The corrected Figure 4 provides a clear visualization of the dramatic shift in the academic landscape over the last five decades.

Corrected Figure 4: Faculty hiring trends at 51 US universities (1964–2014).

Quantitative analysis of the corrected data reveals several distinct phases:

  1. The Pre-Genomics Baseline (1964–1993): For nearly thirty years, hiring in these technical fields remained relatively stagnant, typically averaging between 1 and 3 hires per three-year bin across the 51 surveyed schools.
  2. The Inflection Point (Late 1990s): A significant spike in hiring began in the 1997–1999 bin, where positions jumped to approximately 8. This period coincides with the early stages of high-throughput sequencing technologies.
  3. The Genomic Boom (2000–2011): Hiring continued to climb rapidly, reaching its peak in the 2009–2011 window with approximately 18 hires. This represents a nearly sixfold increase over the historical baseline.
  4. Recent Trends (2012–2014): While there was a slight dip in the most recent bin (approximately 11 hires), the overall level of hiring remains significantly higher than the pre-2000 era, indicating a sustained institutional commitment to computational expertise.

Discussion: Scaling Personnel vs. Scaling Silicon

The "real cost" of sequencing is not merely a matter of purchasing faster processors; it is fundamentally about the "brainpower" required to innovate. The data in Figure 4 confirms that universities recognized this need, aggressively hiring faculty to develop the next generation of bioinformatics tools. The corrected scale demonstrates that the academic response was even more robust than the erroneous figure suggested, with hiring peaking just as sequencing data volumes began their exponential climb.

Limitations

While the corrected figure provides valuable insights, several limitations should be noted:

  • Geographic and Institutional Scope: The dataset is limited to 51 US universities. While these are influential institutions, they may not represent global hiring trends or the significant growth of bioinformatics roles in the private biotechnology sector.
  • Correlation vs. Causality: The authors link this hiring spike to the explosion in sequencing data. However, because the data is sourced from Computer Science departments, it likely reflects a broader "Big Data" and Artificial Intelligence boom that occurred simultaneously across many scientific and engineering disciplines.

Conclusion

This erratum restores the visual integrity of a key piece of evidence regarding the evolution of the computational biology workforce. The corrected Figure 4 accurately portrays the massive scaling of human capital in response to the data challenges of the genomics era, reinforcing the original paper's argument that scaling computation—and the experts who drive it—is essential to keeping pace with data generation.

Summary updated

Abstract

After the publication of this work [1], the authors noticed an error in Fig. 4. The error occurred on the y-axis, instead of running from 0 to 20 it just has 20 at each point. The publisher apologises for the error and any inconvenience caused. The corrected figure is given below: Fig. 4 The number of faculty position hires at 51 US universities in 3-year bins. The recent increase in hiring coincides with the explosion in sequencing data. Data were obtained from http://jeffhuang.com/computer_science_professors.html
BibTeX
@inproceedings{muir2016erratum,
  author = {Paul Muir and Shantao Li and Shaoke Lou and Daifeng Wang and Daniel Spakowicz and Leonidas Salichos and Jing Zhang and George M. Weinstock and Farren J. Isaacs and Joel Rozowsky and Mark Gerstein},
  title = {{Erratum to: The real cost of sequencing: scaling computation to keep pace with data generation}},
  booktitle = {Genome},
  year = {2016},
  doi = {10.1186/s13059-016-0961-9},
}

Loading related content...

Figure