US Patent 7,386,523. K-means clustering using t-test computation. Inventor: Qian Diao. Assignee: Intel.
Protects a method, apparatus, and system for k-means clustering using t-test computation. According to the patent claims, the method performs clustering by detecting similarities and dissimilarities between the components of a dataset representing genes.
US Patent 7,386,399. Method and apparatus for alignment of DNA sequencing data traces. Inventors: Alexandre M. Izmailov, Thomas D. Yager. Assignee: Siemens Healthcare Diagnostics.
Covers a method for aligning nucleic acid data traces that involves selecting reference alignment points from among internal peaks that represent highly conserved bases. “The alignment points may also optionally include the primer peak and/or the full-length peak,” according to the patent abstract. The method assigns reference position numbers to these alignment points that reflect the known relative position of the alignment point, and a sequence position number is assigned to peaks in the data traces “so as to maximize assigning the sequence position number and the reference position number to the same base.” The data traces are then aligned based on the assigned sequence position numbers.
US Patent 7,386,173. Graphical displaying of and pattern recognition in analytical data strings. Inventors: Herbert Kaminski, Jan-Henner Wurmbach, Wolfgang Pusch. Assignee: Bruker Daltonik.
Protects a method for graphically presenting complex analytical data strings containing a multitude of substance-representing peaks, such as mass spectra or chromatograms. The method also encompasses pattern recognition and classification techniques in collections of such data strings. “The invention proposes to highlight, after execution of the pattern recognition or classification algorithms, the significantly participating peaks in the graphical display so that the nature of these peaks, and the substances represented by these peaks, can easily be further investigated,” according to the patent abstract.
US Patent 7,383,237. Computer-aided image analysis. Inventors: Hong Zhang, Garry Carls, Stephen D. Barnhill. Assignee: Health Discovery.
Protects a method for analyzing digitized image data that are input into a processor where “a detection component identifies the areas of particular interest in the image and, by segmentation, separates those objects from the background,” according to the patent abstract. A feature-extraction component then “formulates numerical values relevant to the classification task from the segmented objects.” The results of these steps are input into a trained learning machine classifier, which produces an output that may make a diagnosis, decision, or other action.
US Patent 7,379,824. Computational method and apparatus for predicting polypeptide aggregation or solubility. Inventors: Christopher Dobson, Fabrizio Chiti, Jesus Zurdo. Assignee: Cambridge University Technical Services.
Protects a method for predicting the effect of an amino acid modification on the rate of aggregation of a reference polypeptide. The method calculates the difference in hydrophobicity between the reference polypeptide and a modified polypeptide, calculates the difference in beta-sheet propensity between the reference polypeptide and modified polypeptide, and calculates the difference in charge between the reference polypeptide and modified polypeptide.
US Patent 7,379,822. Protein design automation for protein libraries. Inventors: Bassil I. Dahiyat, Robert J. Hayes, Joerg Bentzien, Klaus M. Fiebig. Assignee: Xencor.
Protects a method that uses protein design automation to generate computationally prescreened secondary libraries of proteins, as well as methods and compositions that use the libraries. According to the patent claims, the method first inputs into a computer a primary library that comprises protein sequences and scaffold protein sequence, and then aligns the protein sequences and scaffold protein sequence to generate an alignment of sequences with variable residue positions. It then analyzes the alignment “to generate a probability distribution of amino acid residues at each variable residue position, such that each variable residue position has a set of possible amino acids,” and recombines the residues from the probability distribution with residues at non-variable residue positions to generate a secondary library of secondary sequences. The method then computationally ranks the secondary library and eliminates unfavorably ranked sequences to generate a tertiary library in which at least one sequence is different from the primary sequences.