HOWARD J. COHEN, Ph.D.
President
Cohen Software Consulting, Inc.
3272 Cowper Street
Palo Alto, CA 94306-3004
(650) 856-8123
(650) 856-4273 (fax)
howard@cohensw.com
http://www.cohensw.com

Selected Projects

• Sequencing Image Analysis
  Worked on portions of a semi-automatic DNA sequencing system, including definitions for extensions to the Standard Gel File format (SGF); work on algorithm design and implementation in bringing a weak DNA signal out of a noisy multi-channel image; work in algorithm design and implementation for lane tracking within that image; and creation of various false-color diagnostic GIF images of the original, signal-processed, and lane tracked data, to enhance understanding of the nature of the data and the workings of the algorithms. Environment was Sun UltraSparc, Solaris, C (gcc and xxgdb), and the gd library for GIF creation.
   
• Chromatogram Archive
  Working on the Chromatogram Archive project, a suite of Perl scripts and C programs to manage and process about 13 million chromatograms of human, mammalian, plant, and pathogen expressed gene sequences, as part of a larger project to reanalyze all of Incyte's proprietary human sequences and a large number of public domain sequences (LifeSeq Gold). This archive includes several terabytes of data. Developed an Oracle database to manage and index this archive as well as make retrieval of specific data fast and simple. Developed a suite of software tools to allow loading of this DB and the assembly and delivery of both small and very large sets of the archived data to customers. Environment is OSF1, Solaris, Oracle Pro*C.
   
• LifeSeq Gold
  Involved in schema design for both in-house production DB and for the DB to be released to customers. Designed and implemented the Annotation program (using public domain databases to understand the assembled proprietary putative genes). Also working on the software development and release environment for multi-platform porting. Environment is Sun Enterprise servers and desktop workstations, DEC Alphas, SGI Octane, Linux and SCO desktops, Perl 5.0, C (gcc and xxgdb), and Oracle, including Pro*C and SQL*Plus. Documentation is in HTML on internal web pages. Development environment is RCS, TCCS, gmake and various scripting languages.
   
• IGP ("The Incyte Genome Project")
  Involved in various aspects of this project to ingest the entire public domain human DNA sequence set as well as the Incyte-proprietary genome sequence data, screen it, and apply gene finding and annotation techniques to it. This project encompasses Incyte's LifeTools database and software with Genomic Enhancements, the LifeSeq Gold data, and the highly efficient distributed processing system described immediately following.
   
• The Brewery and the Farm
  Co-architect and implementor of a system for using coarse-grained parallelism of tasks to distribute them to a farm of client machines (compute servers) of varying sizes, capacities, speeds, and architectures. An Oracle database and a client-side pull manager are the central features of this load-balancing, throughput-enhancing system, essential for the success of LifeSeq Gold and the IGP bimonthly deliveries.
   
• TCCS and Porting
  Installed, set up, and administered TCCS (Trivial Configuration Control System) for a complex multi-user, multi-platform development environment. Developed GNUMakefile's and scripts, user and internal documentation.
   
• Foundation Project
  A next generation database and dataflow architecture for annotation and gene finding in the entire human genome, combining public domain and proprietary data. Participated in the dataflow and database architecture and design. Designed and implemented an XML parser for genomic data and the database loader program, created for speed and efficiency in processing millions of cDNA sequences and gigabytes of gDNA information. Oracle/Pro*C/C, SQL*Plus, HTML documentation.
   
• Mass Spectrometry
  Worked on software for the analysis and display of 2-dimensional Mass Spectrometry data. The project involved decoding and reading the manufacturer's proprietary compressed data format, using a known impurity mass to compute gain corrections, and applying deisotoping and decharging algorithms. Environment was Linux for algorithm development as well as implementation of a high thruput pipeline, g++/C++ and STL; and Windows2000, C++ and STL for the interactive and graphical version of the software.
   
• Patents
  Co-inventor of 114 patents applied for by Incyte Genomics, 4 by Nortel Networks.