Description/summary not set
Other contact details:
Not specified
Interests:
Datamining, algorithms, entrepreneurship, drug discovery, research.
Field/Industry:
Drug Discovery
Occupation/Role(s):
Cheminformatician, Bioinformatician
Organisation(s):
Not specified
Note: some items may not be visible to you, due to viewing permissions.
Contents (click to expand/contract)
Deriving factual information from WHO COVI...
(1)
WHO has posted a CSV that's frequently updated with COVID abstracts. A simple PubMed search reveals thousands of abstracts too. So here's an NLP pipeline to mine these and extract meaningful sentences/facts using Stanford Open Information Extractor and Biological entity tagging. #knime #naturallanguageprocessing #textanalytics #textmining #covid19 #nlp #datascience #WHO
Created: 2020-04-12
Credits:
Insilicoconsulting
Antiviral activity multiclass Predictor
(1)
A simple multi-class antiviral activity predictor based on expanded Pubchem fingerprints and random forest algorithm. Seems to work well to perform quick screens for antiviral molecules.Download training data from CTD (http://ctdbase.org/downloads/) or use this preprocessed version https://www.myexperiment.org/files/2285.html. Write to me at rajeev.gangal@gmail.com regarding any queries.I make no claim that the predictive model is accurate for clinical hypotheses or that it will find mo...
Created: 2020-03-23
| Last updated: 2020-03-25
Credits:
Insilicoconsulting
Genetic Algorithms for BIOISOSTERIC Struct...
(1)
Genetic algorithm workflow to generate novel structures by recombination and mutation and that uses an ML model as a fitness function. Both the fitness function and the fingerprints modified in the model should be the same. ML model cannot have any descriptors other than given fingerprint.Reverse molecular mapping/generation is not part of this workflow and can be done by creating a large Structure-FP database.
Created: 2018-11-06
Credits:
Insilicoconsulting
Texture classification: Knime Image Analytics
(1)
A workflow including image pre-processing and descriptor calculation for building a texture classification model. The workflow can really be used for any kind of image classification task for which has labelled training data.There is significant scope of improvement, both on the image processing , descriptor selection and model building phases..Happy image analytics !
Created: 2015-11-10
Credits:
Insilicoconsulting
Drug-Drug Networks using Matador-DrugBank ...
(1)
These days, Networks of all kinds are all the rage! So here's a trickle contributing to the flood.The workflow generates a network based on Similarity between known drugs based on commonality of their "direct" drug target interactions.The workflow does not provide a comparison between Chemical similarity based network and one based on shared drug targets. It only provides the latter.Other nodes and interfaces to Cytoscape and exported csv can be used to further the network analy...
Created: 2015-09-17
| Last updated: 2015-09-17
Credits:
Insilicoconsulting
SAR by Rgroup decomposition
(1)
This workflow showcases a simple way to use Rgroup decomposition for SAR analysis.Once R-groups are obtained for several sites on the Scaffold, molecular properties like logP, TPSA, number of heavy/hetero-atoms, various kinds of ring counts are calculated.While medicinal chemistry purists may decry calculating properties for fragments/R-groups, correlating these at each site with Ki (or activity) shows up interesting trends.It's immediately apparent if increasing logP at a given site incr...
Created: 2015-09-15
Credits:
Insilicoconsulting