Genetic Code

In the News..

01-10-2006

Chemsilico new addition to product range

Leading-edge predictors built with advanced data selection and neural net techniques utilizing hundreds of topological descriptors, in combination with new proprietary descriptors, are now available. Over 35,000 compounds were used to develop these new validated property predictors.
The Tommi Melon

Chemsilico


CSLogD … Leading-edge Descriptors

Based on our powerful LogP and pKa predictors...

  • Classes of Compounds... Monoprotic, diprotic and 12 classes of ampholytes to coverup to 3 pkas per compound.
  • LogD pH-profiles... In range 0-14 with ion-pairing cutoffs for partition coefficients of ionized species.
  • Leading-edge Descriptors... Uses 158 of our proprietary descriptors and 350 well-known topological descriptors.
  • High External-Validation Accuracy... Q2valid = 0.80 on external testing of 86 drugs with MAE = 0.52
  • Import Results...Directly to ChemFinder™/ISISBase™/MDL QSAR™ for visualization of compounds via CS-SDF output

CSLogD Introduction

CSLogD Experimental

CSLogD Methods & Descriptors


CSLogWS...Exceptional High Accuracy

CSLogWS calculates Intrinsic Aqueous Solubility (LogWSo) and pH Solubility Profiles at pH 2.0, pH 5.0 and pH 7.4.

  • Exceptionally High Accuracy for Intrinsic Solubility..5650 compounds gave a 10% leave out cross validation Q2 = 0.93
  • Solubility-pH Profiles...Aqueous solubility as a function of 0- 3 ionizable groups per compound
  • Fast...Process over 1,500 compounds per minute
  • Leading-edge descriptors...Uses 158 of our proprietary descriptors and 350 well-known topological and E-state descriptors
  • Import Results...Directly to ChemFinder™/ISISBase™/MDL QSAR™ for visualization of compounds via CS-SDF output

CSLogWS Introduction

CSLogWS Experimental

CSLogWS Dataset Profiles

CSLogWS Methods & Descriptors


CSLogP... Fast logP predictor

As an integral member of the ChemSilico group of predictors, CSLogP delivers both accuracy and speed.

  • Exceptionally High Cross-Validation...Cross-validation on approximately 17,000 compounds gave a Q2 = 0.87 with MAE = 0.46 and R2 = 0.89 with MAE = 0.43 for goodness of fit.
  • Fast... Process upwards of 2,000 compounds per minute
  • Import Results...Directly to ChemFinder™/ISISBase™/MDL QSAR™ for visualization of compounds via CS-SDF output

CSLogP Introduction

CSLogP Experimental

CSLogP Dataset Profiles

CSLogP Methods & Descriptors


CSpKa...a new pKa predictor that delivers !!!

  • Exceptional Cross-Validation...Cross-validation Q2s gave 0.78, 0.80, 0.90 and 0.92 on aromatic-N, CO2H, amines and alcohols involving thousands of multiprotic compounds.
  • Ionization Centers...Focuses on 12 protonatable or deprotonatable groups.
  • Impressive External Validition...External validation on 665 new compounds, 312 of which were multiprotic, gave Q2valid = 0.83 with MAE = 0.99
  • Multiple Output...Processes mono to multiprotic compounds, up to five (5) pkas
  • Import Results...Directly to ChemFinder™/ISISBase™/MDL QSAR™ for visualization of compounds via CS-SDF output

CSpKa Introduction

CSpKa Experimental

CSpKa Dataset Profiles

CSpKa Methods & Descriptors


CSBBB...a new leading-edge Blood-Brain Barrier partition predictor

  • Excellent Accuracy by External Validation...External validation gave R2ExVal = 0.62 with MAE = 0.37 on a highly diverse set of of 74 drugs, which is the largest such dataset ever reported with this observed accuracy. The CSBBB QSAR cross-validated, neural net model was based on 103 compounds (Q210% = 0.76, MAE = 0.30) with the use of proprietary topological descriptors.
  • Continuous Processing.. Run upwards of 90,000 compounds per hour to screen any size HTS library
  • Non-fragment Approach.. Nine non-fragment descriptors are applied to highly diverse compounds in the MW range from 101 to 1300 with a LogBB in range from -2 to 2.
  • Import Results...Directly to ChemFinder™/ISISBase™/MDL QSAR™ for visualization of compounds via CS-SDF output

CSBBB Introduction

CSBBB Experimental

CSBBB Dataset Profiles

CSBBB Methods & Descriptors


CSPB ... a new Protein Binding predictor

When a drug binds to proteins in the blood, it can either be directed to its target, or limited in its effectiveness. CSPB can help in designing drugs with low, medium or high binding affinity.

  • Validation Test Results...Cross validation and external test sets give accuracy rates of 91% and 88% respectively, in assigning accurate protein binding activity levels on 345 diverse drug and drug-like compounds
  • Continuous Processing... Process HTS libraries at over 1,500 compounds per minute
  • Output... SDF or CSV to handle large datasets
  • Import Results...Directly to ChemFinder™/ISISBase™/MDL QSAR™ for visualization of compounds via CS-SDF output

CSPB Introduction

CSPB Experimental

CSPB Dataset Profiles

CSPB Methods & Descriptors


CSGenoTox version 2.0...A New Ames Mutagenicity Predictor

  • Exceptionally High External-Validation Accuracy...400 chemically diverse compounds (NCEs) gave an overall accuracy of 84% and 93% concordance on drugs with low false positives (5%) and negatives (3%)
  • Exceptionally High Cross-Validation Accuracy...2963 compounds gave a 10% leave out cross validation accuracy of 89% with low number of false positives ( 3%) and negatives (8%).
  • Leading Edge Descriptors...ChemSilico models are developed from a database of 158 proprietary descriptors along with 350 well-known topological and E-state descriptors
  • Output Results…In SDF, CSV or XLS format to handle large batch processing datasets

CSGenoTox Introduction

CSGenoTox Experimental

CSGenoTox Dataset Profiles

CSGenoTox MI Predictor Comparator

CSGenoTox Methods & Descriptors