Faculty Abstracts
William Shih:
Thesis: “Scalable data storage and autonomous molecular recording are related killer apps for molecular programming”.
Abstract: Scalable DNA storage: DNA has been attracting increasing interest as a media for archival data storage due to the potential for dramatically lower costs and longer lifetimes than conventional tape drives. Towards a practical implementation, what are the challenges for driving down costs of DNA reading and writing by a million-fold from where they are today?
Autonomous molecular recording: Recent advances in biotechnology provide mechanisms for autonomous recording of molecular interactions into DNA snippets. Our ability to generate these snippets scales with the volume of a reaction, while our ability to read these snippets scales with the area of a sequencing array. The implication of this scaling mismatch is that we often will be unable to read directly all of the raw data being recorded. However, molecular computing agents can scale with volume as well. What are examples of molecular computing agents that could pre-process such raw data into useful summaries for subsequent readout by DNA sequencing?
Richard Murray:
Thesis: “Engineers are going to build a self-reproducing artificial cell entirely from scratch in 20 years”
Abstract: The Molecular Programming Project started for me on 30 October 2007 when I sent an email to Niles and Erik asking if they would be interested in “putting something in [to the first-ever NSF Expeditions program call] related to computational aspects of DNA engineering/self-assembly/synthetic biology”. We submitted our proposal on 28 Dec 2007, while Erik was in Madagascar, with Paul and Richard working on the proposal up until the last minute. We were notified of the award on 4 August 2008 and the project officially started on 15 August 2008 (even though we didn’t actually get the grant agreement until 10 October 2008). Motivated by the work we were starting, on 27 July 2009 I gave a talk at the first International Workshop on Bio-design Automation (IWBDA) in which I set a personal target of developing a synthetic cell within 20 years.
In this talk I will summarize where engineers are at on our trek to build a synthetic cell, with approximately 10 years, 10 months and 28 days left to go. I’ll use as a starting point a talk that I gave at the first European Congress on Cell-Free Synthetic Biology on 27 Mar 2017 in which I claimed that (1) We are within 10-15 years of being able to produce genetically-programmed artificial cells and multicellular machines; (2) new biological engineering approaches are required, focusing on moving from thinking about components to think about circuits and systems; and (3) synthetic cells are more promising approach than cell-based machines.
Lulu Qian:
Thesis: “The same molecules will be playing multiple roles in future programmable molecular machines”.
Abstract: Machines are usually made of multiple types of components, each playing a unique role. For example, some roles are structural while others are mechanical or computational. In developing programmable molecular machines, we could design each molecular component to specialize in one type of function and integrate them together to achieve desired system behaviors. Alternatively, we could design the same group of molecules to carry out multiple roles, simultaneously or sequentially. For example, a structural component also performs sensing or information processing at times. A logic component also acts as an actuator. Which of these two strategies will be more widely used in future artificial molecular machines, when they finally approach or exceed the sophistication of the natural ones?
Erik Winfree:
Thesis: “A successful general-purpose molecular programming language won’t look anything at all like conventional programming languages”.
Abstract: Modern programming languages for electronic computers have a number of characteristic features that appear to be essential to their success: well-defined syntax and semantics, a hierarchy of abstractions from high level to low level, transformations between levels and optimizations within a level with well-defined notions of correctness, modular and recursive function calls, deterministic input/output behavior, and so forth. When molecular programming matures, which of these features – if any – will be preserved?
