Bell Eapen MD, PhD.

Plasmin
Proteinases play an important role in conditions associated with inflammatory desquamation of the skin like psoriasis, atopic dermatitis and netherton syndrome. Proteinase inhibitors keep their activity under check. Today I read an article [1] about such an inhibitor called LEKTI (lympho-epithelial Kazal-type-related inhibitor) and the corresponding gene SPINK5, mutations of which is responsible for Netherton syndrome. Authors found that domain 6 and domain 15 inhibit 2 key serine proteinases called hK5 and hK7. They also found that domain 15 with 3 disulphide bonds (and not domain 6) also inhibit plasmin. Plasmin is important in the pathogenesis of pemphigus vulgaris which basically is an autoimmune disease [2]. Hence LEKTI domain 15 may be useful in the treatment of pemphigus vulgaris too. The relevance of the fact that drugs with sulfhydryl groups induce pemphigus with respect to the above finding can be explored.
Interested in further exploring LEKTI bioinformatically?
References:

1. T. Egelrud et al. hK5 and hK7, two serine proteinases abundant in human skin, are inhibited by LEKTI domain 6. British Journal of Dermatology 2005;153:1200-1203.
2. Naito K, Morioka S, Nakajima S, Ogawa H. Proteinase inhibitors block formation of pemphigus acantholysis in experiments models of neonatal mice and skin explants: effects of synthetic and plasma proteinase inhibitors on pemphigus acantholysis. J Invest Dermatol 1989; 93: 173-7.

DSG3 (BJD 2006 Jan 154 pp67-71)
I was always interested in doing some bioinformatics project on basement membrane zone molecules. Authors have studies various DSG3 SNPs and found two different haplotypes in UK and Indian pemphigus vulgaris patients. Authors have suggested further investigation of this gene.
Structure of GPCRs (PLoS Comp Biol Feb 2006 2(2) p 88-99
Feb 2006 PLOS Computational Biology journal has an interesting article about structure prediction of G Protein – Coupled Receptors. Authors have employed the new threading assembly refinement (TASSER) method to predict the structures for all 907 putative GPCRs out of which at least 820 is supposed to have correct folds. The structures are available for noncommercial use from the university website in my links database. It may be useful for my MC1R study too as GRK2 and 6 are GPCR kinases involved in MC1R signaling. Time to refine my protocol further.

UIMA SDK from IBM
Today I explored the UIMA SDK from IBM. It is a software system that analyses large volumes of unstructured information, discover, organize and deliver relevant knowledge. An example they have sited as an application of UIMA is to find potential drug interactions by processing millions of medical abstracts. I will try to explore its application in medical diagnostics and find out whether I can use it in some way for my virtual dermatologist I am working on. Probably it has some applications in bioinformatics too!

In this months biopharm I read about the spiralling cost of biomedical research. New drug approvals have flattened out in recent years, particularly for new molecular entities. As far as man power is concerned, over the last few years there has been an unprecedented increase in the number of bioinformaticians and there is no dearth of lead molecules. However clinical research personal failed to catch up and manufacturers are under pressure to find out more efficient ways to develop medical products. I think very soon the initial phases of clinical studies may become computer simulations and the actual human studies will become more targeted.

Today I read one article about 5 untranslated region or 5UTR which was a new concept for me. The translational efficiency of mRNA depends on various factors like the context of start codon, AUGs within 5 UTR and the structural stability of mRNA. The authors tried to find out the 5UTR features influencing the translational efficiency by comparing various High expression and Low expression mRNAs. They prepared a database out of this information called LEADER_RNA and the translational prediction algorithm.

hi friends,
My name is Abhijit Jadhav.I m doin my graduate project in secondary protein structure prediction using Neural networks.I had few doubts and would be very greatful if you ppl could slove it
1.Is there some benchmark problem for secondary protein structure prediction using neural nets?
2.Is there some neural network solution algorithm for protein structure prediction like RBF networks or SVM?
Thnak you
Abhijit

Today I came across a very interesting home page belonging to Gale Rhodes, Professor of Chemistry, University of Southern Maine. The tutorials on Swiss PBD and Rasmol are interesting and the bioinformatics tutorial for beginners gives a broad overview which is useful even for the experienced. I have posted the link in the resource manager.