William Seffens

Using a neural network to backtranslate amino acid sequences

A neural network (NN) was trained on amino and nucleic acid sequences to test the NN’s ability to predict a nucleic acid sequence given only an amino acid sequence. A multi-layer backpropagation network of one hidden layer with 5 to 9 neurons was used. Different network configurations were used with varying numbers of input neurons to represent amino acids, while a constant representation was used for the output layer representing nucleic acids. In the besttrained network, 93% of the overall bases, 85% of the degenerate bases, and 100% of the fixed bases were correctly predicted from randomly selected test sequences. The training set was composed of 60 human sequences in a window of 10 to 25 codons at the coding sequence start site. Different NN configurations involving the encoding of amino acids under increasing window sizes were evaluated to predict the behavior of the NN with a significantly larger training set. This genetic data analysis effort will assist in understanding human gene structure. Benefits include computational tools that could predict more reliably the backtranslation of amino acid sequences useful for Degenerate PCR cloning, and may assist the identification of human gene coding sequences (CDS) from open reading frames in DNA databases.

Whole transcriptome mRNA secondary structure analysis using distributed computation

We performed a comparative study of Human, Mouse, and Arabidopsis genes to determine if mRNA secondary structure increases with more complex organisms. Calculating the secondary structures of a large number of genes in a transcriptome has a high degree of parallelism, and is suitable to implement on multiple computers. The analysis in this investigation was done on a cluster of eighteen (18) Intel x86- based Windows 2000 workstations. These were linked together as a Network of Workstations (NOW) configuration. A Dell Power Edge 4600 Intel Xeon Windows 2000 dual processor server controlled the NOW workstations and was used for file storage. Single-value means of each transcriptome were calculated as a simple average of all transcripts. This computational research effort is of biomedical interest to provide computational tools to analyze and characterize proteins and mRNAs.

Gene Sequences are Locally Optimized for Global mRNA Folding

An examination of 51 mRNA sequences in GENBANK has revealed that calculated mRNA folding free energies are more negative than expected. Free energy minimization calculations of native mRNA sequences are more negative than randomized mRNA sequences with the same base composition and length. Randomization only of the coding region of genes also yields folding free energies of less negative magnitude than the original native mRNA sequence. Examination of the predicted basepairing within the coding sequence finds an unequal distribution between the three possible frames. The wobble-to-1 frame, which is ”in-frame”, is preferred significantly compared to randomized sets of mRNA sequences. This suggests that evolution may bias or adjust the local selection of codons to favor the global formation of more mRNA structures. This would result in greater negative folding free energies as seen in the 51 mRNAs examined.

Yeast Sage Expression Levels are Related to Calculated mRNA Folding Free Energies

Free energies of folding for native mRNA sequences are more negative than calculated free energies of folding for randomized mRNA sequences with the same mononucleotide base composition and length. Randomization only of the coding region of most genes also yields folding free energies of less negative magnitude than those of the original mRNA sequences. For 79 mRNA sequences selected from a yeast SAGE library, the free energy minimization calculations of native mRNA sequences are also usually more negative than randomized mRNA sequences, as above. This difference can be expressed as a bias using standard deviation units. We also observed that if this yeast SAGE data is grouped according to expression levels, the mean folding free energy bias is different between the high, average, and low expression-level genes. A t-Test for paired two-samples of means shows a significant difference in folding free energies between high and low expression yeast genes. Thus the sequences of these yeast genes typically give rise to more stable secondary mRNA structures in high expression genes than in single-copy genes. The results of this study could serve as a foundation for comparison with other genomes, which in turn will allow investigating how the folding bias may be affected by specific characteristics of each organism, such as growth temperature, dinucleotide composition, or GC content of the genome.

Internet APRS data utilization for biosurveillance applications

APRS is an abbreviation for Automatic Packet Reporting System, and is a method of broadcasting GPS positioning information in real time from packet radio-equipped stations. It was designed in the early 90’s, but it has seen growth in the last few years due to user-friendly software such as WinAPRS or UI-View, and Kenwood’s APRS enabled radio transceivers becoming available. APRS equipped stations send latitude and longitude information, as well as course, speed and altitude of mobile stations. These are commonly set up for use as Search and Rescue operations or special public events such as parades for tactical overviews. Even the International Space Station and a number of low-earth orbiting satellites have an APRS repeater on board, with amateur earth stations watching positions on their PC screens. Many stations also transmit weather data, which is collected for use by the US Weather Service.