Biomedical Engineering Reference
In-Depth Information
Figure 21.8
Example of highlighted alerts on molecules to avoid wasting money on
compounds likely to fail.
pharma customers. In the process we have used it to provide new insights into
the vast amounts of screening data being produced [43] as well as facilitate
global collaborations [19] and provide a means for collaboration [9, 10, 23, 49].
As we see drug discovery become more reliant on networks of collaborators,
we think the need for a cloud-based solution will become dominant.
It is feasible that other cheminformatics software solutions could be pro-
vided to CDD users in the same way, either separately or integrated into the
current platform. For example, it may be of utility to integrate ADME/Tox
models or other quantitative structure-activity relationships (QSARs) [64].
Software developed under the open-source model provides important visibil-
ity into the implementation of descriptors and algorithms, so that computa-
tional chemists can verify the algorithm and suggest or actually contribute
improvements [65]. A number of open-source software packages exist that
calculate descriptors [66, 67] or implement modeling algorithms (e.g., R). Some
groups have used open descriptors and open modeling algorithms to build
QSAR models [67-69] for mutagenicity, cytotoxicity, Caco-2 data, as well as
some drug targets. The data sets used to date have been relatively small. While
there are some toolkits for cheminformatics and bioinformatics [65, 70-72] as
well as proposed Web services [73], no integrated toolkit exists that provides
functionality for end-to-end QSAR training, validation, and prediction. We
have recently used such open-source software with over 100,000 molecules
Search WWH ::
Custom Search