Statistical modeling of biomedical corpora: mining the Caenorhabditis Genetic Center Bibliography for genes related to life span

Background  

The statistical modeling of biomedical corpora could yield integrated, coarse-to-fine views of biological phenomena that complement discoveries made from analysis of molecular sequence and profiling data. Here, the potential of such modeling is demonstrated by examining the 5,225 free-text items in the Caenorhabditis Genetic Center (CGC) Bibliography using techniques from statistical information retrieval. Items in the CGC biomedical text corpus were modeled using the Latent Dirichlet Allocation (LDA) model. LDA is a hierarchical Bayesian model which represents a document as a random mixture over latent topics; each topic is characterized by a distribution over words.     

The shape of human gene family phylogenies

Background  

The shape of phylogenetic trees has been used to make inferences about the evolutionary process by comparing the shapes of actual phylogenies with those expected under simple models of the speciation process. Previous studies have focused on speciation events, but gene duplication is another lineage splitting event, analogous to speciation, and gene loss or deletion is analogous to extinction. Measures of the shape of gene family phylogenies can thus be used to investigate the processes of gene duplication and loss. We make the first systematic attempt to use tree shape to study gene duplication using human gene phylogenies.     

Triple helix formation at (AT)n adjacent to an oligopurine tract.

We have used DNase I footprinting to investigate the recognition of (AT) n tracts in duplex DNA using GT-containing oligonucleotides designed to form alternating G.TA and T.AT triplets. Previous studies have shown that the formation of these complexes is facilitated by anchoring the triplex with a block of adjacent T.AT triplets, i.e. using T11(TG)6to recognize the target A11(AT)6. (AT)6T11. In the present study we have examined how the stability of these complexes is affected by the length of either the T.AT tract or the region of alternating G.TA and T.AT triplets, using oligonucleotides of type T x (TG) y to recognize the sequence A11(AT)11. We find that successful triplex formation at (AT)n (n = 3, 6 or 11) can be achieved with a stabilizing tail of 11xT.AT triplets. The affinity of the third strand increases with the length of the (GT) n tract, suggesting that the alternating G.TA and T.AT triplets are making a positive contribution to stability. These complexes are stabilized by the presence of manganese or a triplex-specific binding ligand. Shorter oligo-nucleotides, such as T7(TG)5, bind less tightly and require the addition of a triplex-binding ligand. T4(GT)5showed no binding under any conditions. Oligo-nucleotides forming a 3'-terminal T.AT are marginally more stable that those with a terminal G.TA. The stability of these complexes was further increased by replacing two of the T.AT triplets in the T n tail region with two C+.GC triplets.     

DNA triple helix formation at target sites containing several pyrimidine interruptions: stabilization by protonated cytosine or 5-(1-propargylamino)dU.

DNase I footprinting has been used to study the formation of parallel triplexes at oligopurine target sequences which are interrupted by pyrimidines at regular intervals. TA interruptions are targeted with third strand oligonucleotides containing guanine, generating G x TA triplets, while CG base pairs are targeted with thymine, forming T x CG triplets. We have attempted to optimize the stability of these complexes by varying the base composition and sequence arrangement of the target sites, and by replacing the third strand thymines with the positively charged analogue 5-(1-propargylamino)dU (U(P)). For the target sequence (AAAT)(5)AA, in which pyrimidines are positioned at every fourth residue, triplex formation with TG-containing oligonucleotides is only detected in the presence of a triplex-binding ligand, though stable triplexes were detected at the target site (AAAAAT)(3)AAAA. Triplex stability at targets containing pyrimidines at every fourth residue is increased by introducing guanines into the duplex repeat unit using the targets (AGAT)(5)AA and (ATGA)(5)AA. In contrast, placing C(+) x GC triplets on the 5'-side of G x TA, using the target (AGTA)(5)TT, produces complexes of lower stability. We have attempted further to increase the stability of these complexes by using the positively charged thymine base analogue U(P), and have shown that (TU(P)TG)(5)TT forms a more stable complex with target (AAAT)(5)AA than the unmodified third strand, generating a footprint in the absence of a triplex-binding ligand. Triplex formation at (AGTA)(5)AA is improved by using the modified oligonucleotide (TCGU(P))(5)TT, generating a complex in which the charged triplets C(+) x GC and U(P) x AT alternate with uncharged triplets. In contrast, placing U(P) x AT triplets adjacent to C(+) x GC, using the third strand oligonucleotide (U(P)CGT)(5)TT, reduces triplex formation, while the third strand with both substitutions, (U(P)CGU(P))(5)TT, produces a complex with intermediate stability. It appears that, although adjacent U(P) x AT triplets form stable triplexes, placing U(P) x AT adjacent to C(+) x GC is unfavorable. Similar results were obtained with fragments containing CG inversions within the oligopurine tract, though triplexes at (AAAAAC)(3)AA were only detected in the presence of a triplex-binding ligand. Placing C(+) x GC on the 5'-side of T x CG triplets also reduces triplex formation, while a 3'-C(+) x GC produces complexes with increased stability.     

A novel series of potent and selective IKK2 inhibitors

A novel series of aminopyrimidine IKK2 inhibitors have been developed which show excellent in vitro inhibition of this enzyme and good selectivity over the IKK1 isoform. The relative potency and selectivity of these compounds has been rationalized using QSAR and structure-based modelling.     

Secondary binding sites for triplex-forming oligonucleotides containing bulges, loops, and mismatches in the third strand

We have used DNase I footprinting to examine the binding of five different 17-mer oligonucleotides to a 53-base oligopurine tract containing four pyrimidine interruptions. Although all the expected triplexes formed with high affinity (K(d) approximately 10-50 nM), one oligonucleotide produced a footprint at a second site with about 20-fold lower affinity. We have explored the nature of this secondary binding site and suggest that it arises when each end of the third strand forms a 7-mer triplex with adjacent regions on the duplex, generating a contiguous 14-base triplex with a bulge in the center of the third strand oligonucleotide. This unusual binding mode was examined by use of oligonucleotides that were designed with the potential to form different length third-strand loops of various base composition. We find that triplexes containing single-base bulges are generally more stable than those with dinucleotide loops, though triplexes can be formed with loops of up to nine thymines, generating complexes with submicromolar dissociation constants. These structures are much more stable than those formed by adding two separate 7-mer oligonucleotides, which do not generate DNase I footprints, though a stable complex is generated when the two halves are covalently joined by a hexa(ethylene glycol) linker. MPE produces less clear footprints, presumably because this cleavage agent binds to triplex DNA, but confirms that the oligonucleotides can bind in unexpected places. These results suggest that extra care needs to be taken when designing long triplex-forming oligonucleotides so as to avoid triplex formation at shorter secondary sites.     

Alternative geometries of DNA looping: an analysis using the SfiI endonuclease

Many processes are governed by proteins that bind to separate sites in DNA and loop out the intervening DNA, but the geometries of the loops have seldom been determined. The SfiI endonuclease cleaves DNA after interacting with two recognition sites, and is a favourable system for the analysis of DNA looping. A gel-shift assay was used here to examine the binding of SfiI to a series of linear DNA molecules containing two SfiI sites separated by 109-170 base-pairs. The complexes in which SfiI trapped a loop by binding to two sites in the same DNA were separated from the complexes containing SfiI bound to separate DNA molecules. Step-wise changes in the inter-site spacing generated two forms of the looped complex with different electrophoretic mobilities. The yields of each looped complex and the complexes from intermolecular synapses all varied cyclically with the inter-site spacing, with similar periodicities ( approximately 10.5 base-pairs) but with different phases. One looped complex predominated whenever the DNA between the sites needed to be underwound in order to produce the correct helical orientation of the binding sites. The other looped complex predominated whenever the intervening DNA needed to be overwound. We conclude that the former has trapped a right-handed loop with a negative node and the latter a left-handed loop with a positive node.     

Quality determination and the repair of poor quality spots in array experiments

Background  

A common feature of microarray experiments is the occurence of missing gene expression data. These missing values occur for a variety of reasons, in particular, because of the filtering of poor quality spots and the removal of undefined values when a logarithmic transformation is applied to negative background-corrected intensities. The efficiency and power of an analysis performed can be substantially reduced by having an incomplete matrix of gene intensities. Additionally, most statistical methods require a complete intensity matrix. Furthermore, biases may be introduced into analyses through missing information on some genes. Thus methods for appropriately replacing (imputing) missing data and/or weighting poor quality spots are required.     

A Taxonomic Search Engine: Federating taxonomic databases using web services

Background  

The taxonomic name of an organism is a key link between different databases that store information on that organism. However, in the absence of a single, comprehensive database of organism names, individual databases lack an easy means of checking the correctness of a name. Furthermore, the same organism may have more than one name, and the same name may apply to more than one organism.