Prediction of Biofilm Inhibiting Peptides: An In silico Approach

Time to predict the biofilm inhibiting peptides using a recent machine learning based tool 'BioFin'.

A recent publication out from our lab. Please explore and write back to (sudheer@iiserb.ac.in) in case

of any problem. Comments are welcome. 

http://journal.frontiersin.org/article/10.3389/fmicb.2016.00949/full

The validated model and other tools developed for the prediction of biofilm inhibiting peptides are available

freely as a webserver at http://metagenomics.iiserb.ac.in/biofin/ and http://metabiosys.iiserb.ac.in/biofin/

ProInflam: a webserver for the prediction of proinflammatory antigenicity of peptides and proteins

Time to predict the proinflammatory antigenicity of peptides and proteins using a recent machine learning based tool 'ProInflam'.

A recent publication out from our lab. Please explore and write back to (sudheer@iiserb.ac.in) in case

of any problem. Comments are welcome. 

http://www.ncbi.nlm.nih.gov/pubmed/27301453

The prediction model for epitope mapping and similarity search are provided as a comprehensive webserver.

Available at http://metagenomics.iiserb.ac.in/proinflam/ and http://metabiosys.iiserb.ac.in/proinflam/

Prediction of peptidoglycan hydrolases- a new class of antibacterial proteins

Time to predict the peptidoglycan hydrolases using a recent machine learning based tool 'HyPe'.

A recent publication out from our lab. Please explore and write back to (ashok@iiserb.ac.in and drsanjivk@gmail.com) in case of any problem. Comments are welcome. 

https://bmcgenomics.biomedcentral.com/articles/10.1186/s12864-016-2753-8

The present tool helps in the identification and classification of novel peptidoglycan hydrolases from complete genomic or metagenomic ORFs. To our knowledge, this is the only tool available for the prediction of peptidoglycan hydrolases from genomic and metagenomic data.

Availability: http://metagenomics.iiserb.ac.in/hype/, http://metabiosys.iiserb.ac.in/hype/.

Metabolomics and drug discovery

Interesting article for those who are interested in the drug discovery using metabolomics approach. They have covered complete pipeline of drug discovery starting from the metaboloic data.

"Emerging applications of metabolomics in drug discovery and precision medicine" 

http://www.nature.com/nrd/journal/vaop/ncurrent/full/nrd.2016.32.html

Xenobiotics metabolism and gut microbes

The role of gut microbes in the metabolism of xenobiotics is known from almost three decades. It is one of the most important field now a days, as the activity and toxicity of xenobiotics is primarily dependent upon the gut microbial profile. The diverse and huge metabolic enzymes of gut microbes are the workhorses of this metabolic activity. The population specific differences in the gut microflora show well correlation with the activity of these gut microbes in modulating the biotransformation of these xenobiotics. 

Recently reports have shown the major impact of the gut microflora on the pharmacological properties of a drugs. Therefore, the gut metagenomic studies focusing on the gut metagenomic potential of these gut microbes will be crucial in the field of precision medicine. For more details you can go through the article published in nature reviews titled as "The microbial pharmacists within us: a metagenomic view of xenobiotic metabolism". The article is freely available at http://www.nature.com/nrmicro/journal/vaop/ncurrent/full/nrmicro.2016.17.html .

Calculation of ROC using R script

Use the following script to calculate ROC and making plot of true positive rate vs false positive rate using your input training data for random forest

data <- read.csv (file = "INPUT training file", sep = ", or \t or  ")
pred = data[,1:20] # Number of columns or features used to describe each sequence
fac = data$Factor # Factor tag for each sequence this can be either positive or negative

library (randomForest)
library (ROCR)
library (pROC)

rf <- randomForest(pred, fac, mtry = 4, ntree = 500, do.trace = 100, na.action = na.fail, importance = TRUE, cv.fold = 10) # use mtry and ntree optimized parameters

OOB.votes <- rf$votes

print ("Area under the curve)
auc(data$Factor, predictions$Pos) #Here Pos is the tag for Positive data

OOB.pred <- OOB.votes[,2]

pred.obj <- prediction (OOB.pred,fac)

RP.perf <- performance(pred.obj, "rec","prec")
plot (RP.perf)

ROC.perf <- performance(pred.obj, "tpr","fpr")
plot (ROC.perf)


For more details please visit
https://cran.r-project.org/web/packages/randomForest/randomForest.pdf
https://cran.r-project.org/web/packages/ROCR/index.html
https://cran.r-project.org/web/packages/pROC/index.html