Johan Kumlien

Recently, there has been much effort in making databases for molecular biology more accessible and interoperable. However, information in text form, such as MEDLINE records, remains a greatly underutilized source of biological information. We have begun a research effort aimed at automatically mapping information from text sources into structured representations, such as knowledge bases. Our approach to this task is to use machine-learning methods to induce routines for extracting facts from text. We describe two learning methods that we have applied to this task--a statistical text classification method, and a relational learning method--and our initial experiments in learning such information-extraction routines. We also present an approach to decreasing the cost of learning information-extraction routines by learning from "weakly" labeled training data.

The human X chromosome is associated with a large number of disease phenotypes, principally because of its unique mode of inheritance that tends to reveal all recessive disorders in males. With the longer term goal of identifying and characterizing most of these genes, we have adopted a chromosome-wide strategy to establish a YAC contig map. We have performed > 3250 inter Alu-PCR product hybridizations to identify overlaps between YAC clones. Positional information associated with many of these YAC clones has been derived from our Reference Library Database and a variety of other public sources. We have constructed a YAC contig map of the X chromosome covering 125 Mb of DNA in 25 contigs and containing 906 YAC clones. These contigs have been verified extensively by FISH and by gel and hybridization fingerprinting techniques. This independently derived map exceeds the coverage of recently reported X chromosome maps built as part of whole-genome YAC maps.