Table 4 Various approaches for predicting the relationship between microbes and diseases
From: In-silico computational approaches to study microbiota impacts on diseases and pharmacotherapy
Category | Method | Description |
|---|---|---|
Path-based methods | KATZHMDA, PBHMDA, MDPH_HMDA, BWNMHMDA, WMGHMDA | Numbers and weighted scores of various sorts of pathways between two nodes are often taken into consideration by path-based approaches |
Random walk methods | RWRHMDA, BiRWHMDA, PRWHMDA, NTSHMDA, BDSILP, BiRWMP, BRWMDA, NBLPIHMDA, RWHMDA | For iterative walking, random walk algorithms provide a graph-based transition probability matrix |
Bipartite local models | LRLSHMDA, NGRHMDA, NCPHMDA, KATZBNRA | BLMs are capable of making independent predictions on both the microbial and disease fronts |
Matrix factorization methods | CMFHMDA, GRNMFHMDA, NMFMDA, KBMF, MDLPHMDA, mHMDA | Matrix factorization methods maximize two latent informative matrices, whose multiplication approximates the association matrix with distinct constraint terms, using different constraint terms |
Network-based methods | MGATMDA, GATMDA, NINIMHMDA, BPNNHMDA,HMDA-PRED, LPHMDA | Because neural networks can adapt to changing input, they can produce the best possible outcome without requiring the output criteria to be redesigned |
Machine learning-based | MDAKRLS | It is a machine learning-based strategy that employs fewer model parameters, saving time and ensuring reliable results |
Other methods | ABHMDA, BMCMDA, MCHMDA | Ensemble learning and matrix completion are two of the most common strategies used |