Biomolecular Engineering

Abstract The protein-folding problem, in its purest form, is too difficult for us to solve in the next several years, but we need structure predictions now. One solution is to try to recognize the similarity between a target protein and one of the thousands of proteins whose structure has been determined experimentally. For very similar proteins, the relationships are easy to find and good models can be built by copying the backbone (and even some sidechains) from the homologous protein of known structure.

ABSTRACT This paper presents two new algorithms for doing mapping from multi-level logic to fieldprogrammable gate arrays. One algorithm, Xmap, is for mapping to table-lookup gates (for example, the Xilinx chip); the other, Amap, is for mapping to selector-based architectures (for example, the Actel chip). Mapping to the Actel architecture can also be achieved by mapping to 3-input tables, and replacing them with equivalent Actel cells (XAmap).

We have three sets of data for testing A set of multiple alignments based on 3D-structure alignment.(Based on FSSP, Z>= 7.0) A diverse set of good-quality protein structures, with no more than 30% residue identity, split into 3 sets for 3-fold cross-validation. Taken from Dunbrack's culledPDB lists, further selected to contain domains in SCOP version 1.55.

Abstract This paper presents a semi-systolic architecture for decoding cyclic linear error-correcting codes at high speed. The architecture implements a variant of Tanner's Algorithm B, modified for simpler and faster implementation. The main features of the architecture are low computational complexity, a simple, regular arrangement of cells for easy layout, short critical paths, and a high clock rate.

Abstract A new method for estimating the packing quality of protein structures is presented. Atoms in high quality protein crystal structures are very uniformly distributed which is difficult to reproduce using structure prediction methods. Packing quality measures can therefore be used to assess structures of low quality and even to refine them. Previous methods mainly use the Voronoi cells of atoms to assess packing quality.

ABSTRACT This paper describes a technique for multi-level logic minimization on functions represented as if-then-else DAGS. We define the concept of Boolean matrices, and give formal definitions of blocks and rectangles and their meanings. We introduce a new heuristic two-column rectangle replacement for finding rectangle coverings of Boolean matrices. This heuristic is well suited for optimizing circuits for area, while controlling the delay. A slight variation of the heuristic optimizes with respect to delay.

Logic synthesis and in particular logic minimization becomes increasingly more important as Very Large Scale Integration (VLSI) continues to o er more and more complexity in a single integrated circuit. As the complexity grows so does the number of di erent design styles. Application speci c intergrated circuits (ASICs) designed from scratch are still very popular, but recently many varieties of programmable logic devices have gained increasing popularity.

Abstract: Supercomputer architectures are not as fast as logic technology allows because memories are slow than the CPU, conditional jumps limit the usefulness of pipelining and prefetching mechanisms, and functional-unit parallelism is limited by the speed of hardware scheduling. We propose a supercomputer architecture called Ring Of Prefetch Elements (ROPE) that attempts to solve the problems of memory latency and conditional jumps without hardware scheduling.

This paper presents a new paradigm for analyzing digital MOS circuits, based on boolean expressions for paths in the switch graph. A new circuit representation, the inverter graph, is described. A new set of electrical design rules are introduced, based on path expressions in the switch graph and cycles in the inverter graph. A program has been written that generates the exclusion constraints implied by the rules.

Tutors NONE YET In the past, I required students to see a writing tutor every other week, but recent budget cuts have reduced the availability of tutors. Many of you still need extra help, so I encourage you to take advantage of whatever tutoring is available. Past experience has been that the best writers were also the most diligent about seeing the tutors (or that the students who used the tutors most ended up writing best—cause and effect are a little hard to separate here).

A critical issue confronting the genome sequencing projects today, and biology in general, is functional and structural characterization of new proteins. This characterization is often inferred by similarity to proteins of known structure or function. Mutation often makes these similarities di cult to detect in distantly related proteins. Finding these evolutionary connections is called remote-homolog detection.

Abstract Bioinformatics is a new engineering field poorly served by traditional engineering curricula. Bioinformatics can be defined in several ways, but the emphasis is always on the use of computer and statistical methods to understand biological data, such as the voluminous data produced by high-throughput biological experimentation including gene sequencing and gene chips.

Page 1. Using back-propagation in local-structure neural networks to do protein design Jes Frellsen, Kevin Karplus Center for Biomolecular Science and Engineering, UC Santa Cruz Introduction We are investigating the use of back-propagation in neural networks to design a primary structure of proteins given a desired secondary structure. We are using neural networks that are trained to predict the secondary structure of proteins from the primary structure.

Abstract: This paper presents a formalization of clocking disciplines used to prevent race conditions in VLSI circuits. A signal-labeling scheme for the two-phase clocking discipline informally described in Mead and Conway [MC] is presented. Rules are given for checking the correct labeling of a circuit consisting of combinatorial logic and memory elements. The signal-labeling conventions are based in part on those of Noice, Mathews, and Newkirk [NMN].

This document discusses our measurement of alignment shift, and two measures that can be derived from alignment shift: mean shift error and shift score. Section 2 presents a precise de nition of shift, including a description of when it can and cannot be measured. In some candidate alignment, shift is a measure of the distance between two pairs of residues aligned in a\ correct" pairwise alignment. The distance is in separation along the protein backbone, as illustrated in Figure 2.1.

Abstract The field of secondary structure prediction methods has advanced again. The best methods now reach levels of 74-76% of the residues correctly predicted in one of the three states helix, strand, or other. In context of the EVA/CASP, we experimented with averaging over the best current methods. The resulting jury decision proved significantly more accurate than the best method. Although the'jury'seemed the best choice on average, for 60% of all proteins one method was better than the jury.

Abstract Introduces Xtmap, a technology mapper for f-input table-lookup cells based on a generate-and-test paradigm. Xtmap can optimize for area and delay simultaneously and produces smaller circuits than previous mappers that considered delay, while matching their delay values. Tables of benchmark results are used to compare Xtmap with Xmap, Dagmap, FlowMap, chortle-d, and mis-pga covering algorithms

Abstract MOTIVATION: A new hidden Markov model method (SAM-T98) for finding remote homologs of protein sequences is described and evaluated. The method begins with a single target sequence and iteratively builds a hidden Markov model (HMM) from the sequence and homologs found using the HMM for database search. SAM-T98 is also used to construct model libraries automatically from sequences in structural databases. METHODS: We evaluate the SAM-T98 method with four datasets.

Each table gives results for one mapper mapping combinational logic to Xilinx 3000 cells. There are three mappers, and results are given both for mapping unoptimized benchmark files and for optimizing with ITEM first, giving six tables in total. The numbers reported are the number of 5-input tables, the number of Xilinx cells (CLBs), the number of levels of logic (unit-delay estimate), a weighted-fanout delay estimate, and the number of gate inputs in the network.

Cornell University, School of Electrical Engineering (also in Computer Science Department) Abstract: Exclusion constraints are boolean equations that must always be satisfied for an MOS circuit to be adequately modeled by simple switch models. The constraints are generated by a new set of electrical design rules, which are simple enough to be checked automatically.