Richard H. Lathrop

Solving the multiple instance problem with axis-parallel rectangles

The multiple instance problem arises in tasks where the training examples are ambiguous: a single example object may have many alternative feature vectors (instances) that describe it, and yet only one of those feature vectors may be responsible for the observed classification of the object. This paper describes and compares three kinds of algorithms that learn axis-parallel rectangles to solve the multiple instance problem. Algorithms that ignore the multiple instance problem perform very poorly. An algorithm that directly confronts the multiple instance problem (by attempting to identify which feature vectors are responsible for the observed classifications) performs best, giving 89% correct predictions on a musk odor prediction task. The paper also illustrates the use of artificial data to debug and compare these algorithms.

A statistical phylogeography of influenza A H5N1.

The geographic diffusion of highly pathogenic influenza A H5N1 has largely been traced from the perspective of the viruss victims. Birds of a variety of avian orders have been sampled across localities, and their infection has been identified by a general genetic test. Another approach tracks the migration from the perspective of the virus alone, by way of a phylogeography of H5N1 genetic sequences. Although several phylogenies in the literature have labeled H5N1 clades by geographic region, none has analytically inferred the history of the viruss migration. With a statistical phylogeography of 192 hemagglutinin and neuraminidase isolates, we show that the Chinese province of Guangdong is the source of multiple H5N1 strains spreading at both regional and international scales. In contrast, Indochina appears to be a regional sink, at the same time demonstrating bidirectional dispersal among localities within the region. An evolutionary trace of HA1 across the phylogeography suggests a mechanism by which H5N1 is able to infect repeated cycles of host species across localities, regardless of the host species first infected in each locale. The trace also hypothesizes amino acid replacements that preceded the first recorded outbreak of pathogenic H5N1 in Hong Kong, 1997.

Computational identification of a transiently open L1/S3 pocket for reactivation of mutant p53

The tumour suppressor p53 is the most frequently mutated gene in human cancer. Reactivation of mutant p53 by small molecules is an exciting potential cancer therapy. Although several compounds restore wild-type function to mutant p53, their binding sites and mechanisms of action are elusive. Here computational methods identify a transiently open binding pocket between loop L1 and sheet S3 of the p53 core domain. Mutation of residue Cys124, located at the centre of the pocket, abolishes p53 reactivation of mutant R175H by PRIMA-1, a known reactivation compound. Ensemble-based virtual screening against this newly revealed pocket selects stictic acid as a potential p53 reactivation compound. In human osteosarcoma cells, stictic acid exhibits dose-dependent reactivation of p21 expression for mutant R175H more strongly than does PRIMA-1. These results indicate the L1/S3 pocket as a target for pharmaceutical reactivation of p53 mutants.

Compass: A shape-based machine learning tool for drug design

SummaryBuilding predictive models for iterative drug design in the absence of a known target protein structure is an important challenge. We present a novel technique, Compass, that removes a major obstacle to accurate prediction by automatically selecting conformations and alignments of molecules without the benefit of a characterized active site. The technique combines explicit representation of molecular shape with neural network learning methods to produce highly predictive models, even across chemically distinct classes of molecules. We apply the method to predicting human perception of musk odor and show how the resulting models can provide graphical guidance for chemical modifications.

ARIADNE: pattern-directed inference and hierarchical abstraction in protein structure recognition

The macro-molecular structural conformations of proteins exhibit higher order regularities whose recognition is complicated by many factors. ARIADNE searches for similarities between structural descriptors and hypothesized protein structure at levels more abstract than the primary sequence.

Information‐theoretic dissection of pairwise contact potentials

Pairwise contact potentials have a long, successful history in protein structure prediction. They provide an easily‐estimated representation of many attributes of protein structures, such as the hydrophobic effect. In order to improve on existing potentials, one should develop a clear understanding of precisely what information they convey. Here, using mutual information, we quantified the information in amino acid potentials, and the importance of hydropathy, charge, disulfide bonding, and burial. Sampling error in mutual information was controlled for by estimating how much information cannot be attributed to sampling bias. We found the information in amino acid contacts to be modest: 0.04 bits per contact. Of that, only 0.01 bits of information could not be attributed to hydropathy, charge, disulfide bonding, or burial. Proteins 2002;49:7–14.

Current Limitations to Protein Threading Approaches

A short review of the threading approach to protein structure prediction, including presentation of some open statistical problems. Also discussed is one of the likely sources of the current limited success, that being the form of the pairwise potentials used in most threading approaches.

Predicting Protein Structure With Probabilistic Models

Publisher Summary This chapter discusses a probabilistic approach to various algorithmic structure-prediction problems. Important information about protein structure that is difficult or impossible to handle using standard statistical prediction algorithms, such as circular-dichroism studies and disulfide-bridge mapping, are modeled in the chapter as part of the mathematical formulation. The resulting prediction algorithms compute probabilities about the protein structure, given the amino-acid sequence and other experimental results. These probabilities measure the support for modeled hypotheses and are based on the particular protein sequence being analyzed. The probabilistic formulation involves two kinds of quantities—observed and unobserved. The chapter discusses the use of observed quantities to draw inferences about unobserved quantities, such as secondary structure, active sites, or amino-acid interactions with each other, with the solvent, or with ligands. It associates a measure of confidence with each inference.

HB tag modules for PCR-based gene tagging and tandem affinity purification in Saccharomyces cerevisiae

We have recently developed the HB tag as a useful tool for tandem‐affinity purification under native as well as fully denaturing conditions. The HB tag and its derivatives consist of a hexahistidine tag and a bacterially‐derived in vivo biotinylation signal peptide, which support sequential purification by Ni2+‐chelate chromatography and binding to immobilized streptavidin. To facilitate tagging of budding yeast proteins with HB tags, we have created a series of plasmids with various selectable markers. These plasmids allow single‐step PCR‐based tagging and expression under control of the endogenous promoters or the inducible GAL1 promoter. HB tagging of several budding yeast ORFs demonstrated efficient biotinylation of the HB tag in vivo by endogenous yeast biotin ligases. No adverse effects of the HB tag on protein function were observed. The HB tagging plasmids presented here are related to previously reported epitope‐tagging plasmids, allowing PCR‐based tagging with the same locus‐specific primer sets that are used for other widely used epitope‐tagging strategies. The Sequences for the described plasmids were submitted to GenBank under Accession Numbers

Predicting Positive p53 Cancer Rescue Regions Using Most Informative Positive (MIP) Active Learning

Many protein engineering problems involve finding mutations that produce proteins with a particular function. Computational active learning is an attractive approach to discover desired biological activities. Traditional active learning techniques have been optimized to iteratively improve classifier accuracy, not to quickly discover biologically significant results. We report here a novel active learning technique, Most Informative Positive (MIP), which is tailored to biological problems because it seeks novel and informative positive results. MIP active learning differs from traditional active learning methods in two ways: (1) it preferentially seeks Positive (functionally active) examples; and (2) it may be effectively extended to select gene regions suitable for high throughput combinatorial mutagenesis. We applied MIP to discover mutations in the tumor suppressor protein p53 that reactivate mutated p53 found in human cancers. This is an important biomedical goal because p53 mutants have been implicated in half of all human cancers, and restoring active p53 in tumors leads to tumor regression. MIP found Positive (cancer rescue) p53 mutants in silico using 33% fewer experiments than traditional non-MIP active learning, with only a minor decrease in classifier accuracy. Applying MIP to in vivo experimentation yielded immediate Positive results. Ten different p53 mutations found in human cancers were paired in silico with all possible single amino acid rescue mutations, from which MIP was used to select a Positive Region predicted to be enriched for p53 cancer rescue mutants. In vivo assays showed that the predicted Positive Region: (1) had significantly more (p<0.01) new strong cancer rescue mutants than control regions (Negative, and non-MIP active learning); (2) had slightly more new strong cancer rescue mutants than an Expert region selected for purely biological considerations; and (3) rescued for the first time the previously unrescuable p53 cancer mutant P152L.