NDIST'13 (Day 3)
Day three continues with interesting talks on biological systems.
Warren Hunt - Phylogeny
- Working on phylogeny issues. Building classifications of tree
structured data.
- TASPI - operates on a classical Scheme tree structure built with
lists.
- Common sub-tree reduction to reduce size[^0]
[^0]: This feels like a classic BDD generation issue
- Every time biologists talk to him they:
- Don’t know how much data they have
- Don’t know how it is structured
- Discussed working from the bottom of the tree up
- Computer scientists would prefer working from the bottom
Rebuild trees by knowing the occurrence rate of subtrees.
Specifically, know that a tree is constructed from subtrees and
specify those occurrences.
- Bio-Machine encoding scheme
- Quickly generate trees
- Mechanical proof possible (associativity, commutativity, and what
not)
- Possible method for programmable and generative simulation
algorithms
Jacob Beal - Workflow talk
Discussed the concept of a biocompiler that translates “operators” to
“motifs” which is pretty interesting.
- Using compilation optimizations to make biological implementations
more efficient
Panel Session - Defenses, Biological and Computational
Can we learn from one another. Can we avoid repeated mistakes.
Howie Schrobe
Computation
- Near perfect components
- Core design formed in an era of scarcity
- Core design formed in isolated environment
- Evolutionary pressure from market - price performance
- Self-regulation and adaptation rarely considered - Runs open-loop
- No enterprise-wide survivability mechanisms
Biological
- Failable components
- Abundance of resources
- Evolution in ecosystem of predators and parasites
- Evolutionary pressure from ecosystem - survivability
- Self-regulation and adaptation are core mechanisms. Closed loop
control.
- Diversity for population survival. Public-health systems in human
society
Tom Knight - Biological Protection Mechanisms
- Biology
- Replicate cell or system to avoid single points of failure
- Biology is in the business of making more copies of itself
- Survival and competition is the fundamental strategy
- Survival of single cells are not necessary
- Evolution plays a central role
- DNA Repair
- Damage from UV chemicals
- Base repair
- Damaged single bases
- Repair by matching to opposite strand
- Identify parent strand with methylation marks
- Strand breaks
- Homologous end repair - Needs a “matching sequence” on the
other end
- Non-homologous end repair - Highly error prone, desperation move
- DNA Invasion
- Restriction enzymes
- Protect species DNA with unique marks - methylation of specific
sequences
- Attack and destroy DNA not marked
- Works within all cells of a specific species
- CRISPERs
- Find representative sequences of an attacker
- Incorporate these sequences into a list of “likely targets”
- Selectively attack these sequences when found
- Works at the genetic level (memory across generations)
-
Change the Code
- Chemical Attacks (Toxins)
- Pump the chemical out of the cell
- Actively modify the chemical before it can do harm
- Modify your chemistry to be immune to toxin - Change in protein
sequence or turn on alternative pathways
- Inactive sate
- Small fractions of cells are active at any time
- Multi-cellular immune system
- Innate immunity
- Present from birth
- Detects common pathogen characteristics
- General rather than specific
- Fast
- Acquire immunity
- Slow, learned response
- Very effective after initial exposure
- Adaptive Immunity
- Initial targeting by B cells with poor specificity
- High speed mutation of activated B cells
- Selection of higher specificity binders
- Long term memory of these binders
- Targets include not just pathogens, but also cell internals
- Presentation of short peptides by all cells in MHC
- Comparison to learned “self”
- Detection of e.g. cancers, mutations
- Programmed cell death
Howie Shrobe (again)
At least 20-30% of the body’s resources are involved in constant
surveillance and containment.
Part of the innate immune system engages the adaptive immune system
Adaptive immunity
- Hardware analog of immune system detects anomaly
- Software system analog of adaptive immune system is signaled
- System model is used to perform diagnosis
- System model is adapted with new attack-specific detector
- Adaptive immune system synthesizes plan to get around problem and
patch to remove specific vulnerability
Workforce & Degradation is the Economic Trade-off
Processes of attacker and defender
Attacker:
- Recon
- Penetration
- Escalation
- Exploitation
Defender:
- Detection
- Diagnosis
- Remediation
- Regeneration
Originally the subsystem that was exploited by the Car Shark system
was isolated from the real-time control systems. But, to make
diagnostics and only updating work, these were later bridged.
Bingo!
Stuxnet attacked a system that was air-gapped.
Pat Lincoln -
Premises from 2000 are still valid
- We can not control the distribution of technology and information
enabling the manipulation of biological systems
- Ineffective attempts to forbid access to basic biology manipulation
incur high cost
- Delayed gains
- Missed opportunities
- Threats could arise form nature, nation states, loosely organized
groups, and individual
-
Threats could target any part of the living world relevant to human
welfare
- Biosafety
- What do you protect
- Humans
- Agriculture
- Ecology of niches
- Gaia
- From what do you protect it
- Organisms from elsewhere and other times
- Unintended consequences of organisms and parts designed by humans
- Organisms and parts that are newly synthesized
-
Biosecurity - Security against humans doing bad things with biotech.
-
Mitigation of risks:
- Speed
- Cultural
- Legal
- Technical
- Telomere-like generation counters
- Alien-like overlapping code regions
- Random parity-like signature checks