- Computational Biology
- Bio-molecular Recognition
- Adaptive Sensor Design
- Information Theory
- Machine Learning
- Femto-second laser micro-machining
- Quantum Cryptography
Recent discoveries enabled through Next Generation Sequencing (NGS) and Genome-Wide Association Studies (GWAS) show astounding correlations between single-point mutations in our genetic code and disease phenotypes. Thus, the ability to precisely add, delete and edit specific bits of DNA is the catalyst that will drive the next generation of gene therapeutics. Yet, current genome-engineering tools involve either complex redesign of a new protein for each new target sequence (Zinc-fingers) or have high off-target activity (TALENs, CRISPR) leading to spurious cuts at sites different than the one targeted and unpredictable mutations.
On the other hand, nature-engineered DNA cutting enzymes called endonucleases, i.e. proteins that bacteria use to destroy viruses (bacteriophages) by cleaving the foreign DNA, have very high sequence specificity with minimal off-target cleavage, which renders them the ideal DNA cutting tools. Endonucleases though have a significant shortcoming which is that they are currently limited to those found in biology with no general means for programming them to cut new DNA sequences of interest. The long term goal of this collaborative project between Molecular Machines group at MIT Media Lab, Systems Biomedicine group at Bar Ilan university and Bio-molecular design group at Weizmann Institute, is to be able to computationally design an endonuclease that will cut a given arbitrary DNA sequence with high-precision.
Using a novel information theoretic approach to tackle the odorant recognition process happening in our olfactory receptors. Using this approach and by doing a clever weighted sequence logo to “show” the hidden information in olfactory receptors’ protein sequences, I had a fantastic result, pinning down only 2 (of 308) odorant recognition important residues. The result is consistent with the results of molecular dynamics analysis, yet in our approach we used only information theory and no complicated and time prodigal biophysics simulations. Thus, I tested the hypothesis that information theoretic concepts can be usefully adapted to complement molecular dynamics simulation and thermodynamics for analyzing molecular mechanisms of life, from transcription and translation, to immune response and receptor protein ligand recognition, let alone the function of complex systems such as the brain. Quantum and classical information theory as applied to biosystems has started yielding surprising insights that coupled to advances in information technology are now open to experimental testing.
Abstract:Neurofeedback (NFB) uses electro-encephalogram (EEG)- coupled biofeedback to reinforce desired brainwave patterns, usually defined in comparison to a normative database of healthy controls. Although NFB shows great promise for enhancing performance and treating clinical disorders, proof of efficacy has been limited, in part due to individual vari- ation. Here we characterize a novel self-calibrating protocol (SCP) method coupled with five standard machine learning algorithms to classify brain states corresponding to the ex- perience of “pain” or “no pain”. Our results indicate that commercially available, wearable EEG sensors provide suf- ficient data fidelity to robustly differentiate the two “oppo- site” brain states. We describe first steps towards a versatile and reliable platform for individualized EEG neurofeedback, bypassing the pitfalls of using “normed” neurophysiological states, and paving the way towards personalized therapies and brain training that is self-calibrated to arbitrary brain states corresponding to a wide variety of individual needs.
Keywords: EEG, neurofeedback, biofeedback, self-calibrating protocols, wearable
Abstract: This work surveys the implementation of a densly packed micro-electrode array housed in a microfluidic chamber. The chamber can facilitate the growth of a cerebral organoid and monitor it’s neural ac- tivity, in principle, with single neuron temporal and spatial resolution. The proposed way of implementation includes femtosecond laser micro- machining and metallization of PDMS (polydimethylsiloxane) and glass for the construction of the micro-electrodes in a lithographic rather than soft-lithographic way.
Keywords:micro-electrode array, microfluidics, MMEA, brain organoid, femtosecond laser, micro-machining, metallization, organ-on-a-chip
Abstract: Fingerprints provide an elegant and cost-effective solution to the Equality Problem in communi- cation complexity. Their quantum counterpart is one example where an exponential gap exists between classical and quantum communication cost. Moreover, recent publications have pro- posed efficient ways to construct and work with quantum fingerprints in practice. Apart from the savings in communication cost, quantum fingerprints have an additional, inherent feature, namely the ability to hide information, which renders them a perfect candidate for Quantum Cryptography. This thesis reviews quantum fingerprints both as a communication complexity asset as well as a crypto-primitive and investigates the use of Quantum Fingerprinting to im- plement experimentally feasible Quantum Money schemes. We propose a public-key Quantum Money scheme comprising Quantum Fingerprints as well as an experimental implementation of it, feasible with current technology.
Keywords: Quantum Cryptography, Quantum Fingerprints, Quantum Optics, Quantum Money, Time-bin Qubits