Unleashed Informatics Launches SMIDSuite Subscription Service

Unleashed Informatics Limited has announced the launch of SMIDSuite; a small molecule information suite containing the Small Molecule Interaction Database (SMID), SMIDBlast and SMID Genomes.

"Computational drug screening has suffered from irrational exuberance," says Eric Andrade, CEO of Unleashed Informatics.

"Too many false positives are returned in an effort to generate quantity in the in-silico predictions. Scientists are left wanting for quality."

"SMIDSuite is a significant improvement over other approaches by delivering a short-list of highly probable small molecule ligands ready for experimentation, with each ligand ranked by a verified measure of confidence."

At the heart of SMIDSuite lies the Small Molecule Interaction Database (SMID).  SMID is an expanding database of small molecule - protein domain interactions derived from Protein Data Bank 3D structure records containing over 230,000 experimentally observed domain/small molecule interactions. 

By narrowing down from protein-ligand to domain-ligand interactions, SMID is able to cluster similar interactions and demonstrate subtle binding patterns that are not otherwise obvious, particularly in the case of human proteins where proteins may be alternatively spliced.

Unleashed Informatics is offering free, unlimited access to the Small Molecule Interaction Database (SMID) browser and FTP site to registered academic customers.

SMIDBlast builds upon SMID’s foundation interaction data to offer a better way to predict small molecule binding sites on query sequences. 

SMIDBlast runs NCBI's sensitive Reverse-PSI BLAST program enabling customers to match a protein sequence of interest against a growing library of over 3,000 domain definitions with over 5,000 biologically significant small molecules to predict binding sites on the query protein.

SMIDBlast displays hits for which interaction records exist in SMID, and highlights amino acids predicted to interact with small molecules based on interactions observed on similar conserved domains found on 3D structures. 

SMIDBlast also provides a summary list of small molecules predicted to bind the query grouped into non-overlapping binding sites on the query sequence.

"The ‘N1’ designation in the avian flu virus, H5N1, stands for ‘neuramidase1’, and part of the panic over a possible flu pandemic arises from mutations reported to make it resistant to Tamiflu^® and Relenza^®," according to Dr. Christopher Hogue, Chief Scientific Officer of Unleashed Informatics.

"An undergraduate student at the University of Waterloo used SMIDBlast on this mutant protein target to retrieve 5 high scoring small molecules in seconds," explained Dr. Hogue.

He adds, "Then with further drug-docking to a model of the bird flu neuramidase, she showed that compounds published by academics in 1998 (PMID 10547289) and 1999 (PMID 10547289) outscored conventional cures."

SMIDBlast is designed to protect users from over-prediction and false-positives by accounting for factors neglected by other similar systems.

These factors include - Transitive alignment errors, non-biological small molecules, crystallographic condition artifacts, and poor assumptions regarding ion binding sites, incorrect treatment of complex carbohydrates.

In addition, SMIDBlast features improved sequence search sensitivity and specificity, binding site overlap and residue information, and a means to rank small molecule results that reflects known protein-ligand preferences.

Unleashed Informatics offers the SMIDBlast web application freely to registered academic and commercial users for up to three queries per month. Additional usage requires a SMIDSuite license.

SMIDGenomes is a database containing over 12 million predicted small molecule interactions spanning nearly 1,600 complete genomes. SMIDGenomes annotates the small molecule binding sites of 50% of human RefSeq proteins.

SMIDGenomes offers an interface to browse, search or compare predicted small molecule interactions in an 'organism-specific' or cross-genomes manner.

SMID Genomes is built by running SMID-BLAST over the NCBI non-redundant (NR) sequence database, with genome information obtained from the NCBI's RefSeq database. 

The overlap of small molecule hits for up to 5 different organisms can be compared to study potential drug targets, drug cross-reactivity against all species listed, and specific differences in small molecules bound by different strains such as pathogens causing Tuberculosis or Anthrax.

"We knew we got it right when bacterial host-pathogen comparisons lit up with significant known true positive, broad-spectrum antibiotics," says Dr. Hogue. 

"For Borrellia burgdorferi, the Lyme Disease pathogen - a very difficult organism in which to create knock-outs - we came up with a short list that eliminated 98% of the knock-out work required."

"That left only 31 genes as targets unique to the pathogen. The binding sites on these targets are not found in the host human genome - and so are more likely to lead to a cure with no side effects showing up late in clinical trials."

Each SMID Genomes record is based on a high-quality alignment to a 3D structure with mapped binding site information for each small molecule site on the query amino acid sequence. 

Such SMID Genomes content represents the basis for a predicted chemical interaction BIND record. 

"Another use for this data is to compare genomic SNP sites to potential small molecule interactors in genetic diseases," adds Dr. Hogue

"As well as pathway and interaction network analysis to find chemical genetics control points."

Unleashed Informatics offers free, unlimited access to one SMIDGenome sample organism - Saccharomyces cerevisia. Unlimited access to data relating to additional organisms requires a SMIDSuite license.