NetInfer @ LMMD

NetInfer is a command-line toolkit written in C++ programming language, which can be used for predicting potential targets for approved drugs, drug candidates failed in clinical trials, and new chemical entities. It is light-weight and does not need the support of any third-party math libraries such as linear algebra libraries. To accelerate the calculation and decrease the cost of memory space, different data structures were designed for sparse and dense matrices. NetInfer provides various functions, such as link prediction, 10-fold cross validation, leave-one-cross validation, and external validation. In a uniform platform, researchers can input their in-house data and then obtain predictive lists or evaluation indicators. Several our previous methods were implemented in this toolkit, including:
  • Network-Based Inference (NBI)
  • Substructure-Drug-Target Network-Based Inference (SDTNBI)
  • Balanced Substructure-Drug-Target Network-Based Inference (bSDTNBI)
Download:

The latest versions are available upon request at now:

Updated date Platform Comment
2017-2-1964-bit LinuxCompiled by GCC 5.3.0 on CentOS 5.8
Input file format:
Tab-delimited text files can be used as input files for our toolkit. Each line in a input file represents one linkage between two nodes A and B, containing 5 parts:
  1. the type of node A (e.g. DRUG or COMPOUND)
  2. the identifier of node A (e.g. DB00945)
  3. the type of node B (e.g. TARGET)
  4. the identifier of node B (e.g. P23219)
  5. the weighted value of the edge between A and B (e.g. 1.0)
References (#Co-first authors, *Corresponding authors):
  1. Wu ZR, Cheng FX*, Li J, Li WH, Liu GX, Tang Y*. SDTNBI: an integrated network and chemoinformatics tool for systematic prediction of drug-target interactions and drug repositioning. Brief Bioinform, 2017, 18(2): 333-347. DOI: 10.1093/bib/bbw012.
  2. Wu ZR#, Lu WQ#, Wu D, Luo AQ, Bian HP, Li J, Li WH, Liu GX, Huang J*, Cheng FX*, Tang Y*. In silico prediction of chemical mechanism of action via an improved network-based inference method. Br J Pharmacol, 2016, 173(23): 3372-3385. DOI: 10.1111/bph.13629.
  3. Fang JS#, Wu ZR#, Cai CP, Wang Q, Tang Y*, Cheng FX*. Quantitative and systems pharmacology. 1. In silico prediction of drug–target interactions of natural products enables new targeted cancer therapy. J Chem Inf Model, 2017, DOI: 10.1021/acs.jcim.7b00216.
  4. Wu ZR#, Lu WQ#, Yu WW, Wang TDY, Li WH, Liu GX, Zhang HK, Pang XF, Huang J, Liu MY*, Cheng FX*, Tang Y*. Quantitative and systems pharmacology 2. In silico polypharmacology of G protein-coupled receptor ligands via network-based approaches. Accepted.

Tutorials

1. Introduction:

The schematic diagram of substructure-drug-target network-based inference (SDTNBI):


Reference:
Wu ZR, Cheng FX*, Li J, Li WH, Liu GX, Tang Y*. SDTNBI: an integrated network and chemoinformatics tool for systematic prediction of drug-target interactions and drug repositioning. Brief Bioinform, 2017, 18(2): 333-347. DOI: 10.1093/bib/bbw012.

2. Benchmark datasets:

Following files were used in our study of SDTNBI, you can use some of them to try our tool:

3. Newly predicted drug-target interactions:

We predicted thousands of new potential drug-target interactions for FDA approved or clinical investigational drugs via SDTNBI, which are freely available for future validation for experimental scientists.

4. Prioritizing targets for known drugs:

An usage example:
netinfer -method nbi -nbi_k 2 -command predict -node_type DRUG TARGET -length 20 -training_set GPCR_DT.tsv+GPCR_DS_MACCS.tsv -output GPCR_DT_OUT.tsv

Parameters:

-nbi_k <K> The number of resource spreading processes (k).
-command predict To execute the command of prediction.
-node_type <A> <B> To predict potential linkages between A (drug) and B (target).
-length <N> To predict N potential drug-target interactions for each drug.
-training_set <DT>+<DS> The input files of drug-target interactions and drug-substructure linkages.
-output <OUT> The output file of newly predicted drug-target interactions.

5. Predicting targets for new chemical entities:

An usage example:
netinfer -method nbi -nbi_k 2 -command predict -node_type COMPOUND TARGET -length 20 -training_set GPCR_DT.tsv+GPCR_DS_MACCS.tsv+GPCR_CS_MACCS.tsv -output GPCR_CT_OUT.tsv

Parameters:

-nbi_k <K> The number of resource spreading processes (k).
-command predict To execute the command of prediction.
-node_type <A> <B> To predict potential linkages between A (new chemical entity) and B (target).
-length <N> To predict N potential new chemical entity-target interactions for each new chemical entity.
-training_set <DT>+<DS> <CS> The input files of drug-target interactions, drug-substructure and new chemical entity-substructure linkages.
-output <OUT> The output file of new predicted new chemical entity-target interactions.

6. Executing 10-fold cross validation:

An usage example:
netinfer -method nbi -nbi_k 2 -command 10fcv -node_type DRUG TARGET -length 5 10 15 20 -repeat 3 -random_seed 12345 -training_set GPCR_DT.tsv+GPCR_DS_MACCS.tsv -output CV_OUT.tsv CV_ROC.tsv CV_PR.tsv

Parameters:

-nbi_k <K> The number of resource spreading processes (k).
-command 10fcv To execute the command of 10-fold cross validation.
-node_type <A> <B> To do 10-fold cross validation between A (drug) and B (target).
-length <L1> <L2> <L3> ... To calculate evaluation indicators under different L values.
-repeat <M> ... The number of cross validation.
-random_seed <SEED> ... The random seed used for dividing the dataset into training set and test set.
-training_set <DT>+<DS> The input files of drug-target interactions and drug-substructure linkages.
-output <OUT> <ROC> <PR> The output files of evalution indicators, ROC curve and precision-recall curve.

7. Executing QSAR-style leave-one-out cross validation:

An usage example:
netinfer -method nbi -nbi_k 2 -command loocv -node_type DRUG TARGET -length 5 10 15 20 -training_set GPCR_DT.tsv+GPCR_DS_MACCS.tsv -output CV_OUT.tsv CV_ROC.tsv CV_PR.tsv

Parameters:

-nbi_k <K> The number of resource spreading processes (k).
-command loocv To execute the command of leave-one-out cross validation.
-node_type <A> <B> To do leave-one-out cross validation between A (drug) and B (target).
-length <L1> <L2> <L3> ... To calculate evaluation indicators under different L values.
-training_set <DT>+<DS> The input files of drug-target interactions and drug-substructure linkages.
-output <OUT> <ROC> <PR> The output files of evalution indicators, ROC curve and precision-recall curve.

8. Executing external validation:

An usage example:
netinfer -method nbi -nbi_k 2 -command ev -node_type COMPOUND TARGET -length 5 10 15 20 -training_set GPCR_DT.tsv+GPCR_DS_MACCS.tsv+GPCR_CS_MACCS.tsv -test_set GPCR_CT.tsv -output EV_OUT.tsv EV_ROC.tsv EV_PR.tsv

Parameters:

-nbi_k <K> The number of resource spreading processes (k).
-command ev To execute the command of external validation.
-node_type <A> <B> To do external validation between A (new chemical entity) and B (target).
-length <L1> <L2> <L3> ... To calculate evaluation indicators under different L values.
-training_set <DT>+<DS> <CS> The input files of drug-target interactions, drug-substructure and new chemical entity-substructure linkages.
-test_set <CT> The input file of external validation set (new chemical entity-target interactions).
-output <OUT> <ROC> <PR> The output files of evalution indicators, ROC curve and precision-recall curve.


Page last updated at 2017-11-08 09:11:36 (Asia/Shanghai)

Copyright © 2015-2017 Laboratory of Molecular Modeling and Design, Shanghai Key Laboratory of New Drug Design,
School of Pharmacy, East China University of Science and Technology. All rights reserved.