The Support Vector Machine (SVM) Calculator accepts gene expression values from log2 normalized microarray data and copy number values as integers. It can also accept qRT-PCR data, but equal width binning is necessary for comparison to the training set (paclitaxel and gemcitabine models only). SVMs were trained for paclitaxel and gemcitabine in Dorman et al. SVMs for doxorubicin, epirubicin, methotrexate, tamoxifen and 5-fluorouracil were built in Mucaki et al. (2016), and for cisplatin, carboplatin and oxaliplatin in Mucaki et al. (2019). SVMs for 6 tyrosine kinase inhibitors (Erlotinib, Gefitinib, Imatinib, Lapatinib, Sorafenib and Sunitinib) are now available.
After you input expression and/or copy number data, the calculator will return the score - the distance from the SVM hyperplane - and the prediction for whether the cell line or patient is sensitive or resistant. Models were developed by normalizing across both training and testing (patient) datasets. This calculator does not provide such a normalization because it makes predictions on individual patients. Nevertheless, we have verified that the expression differences seen between genes and each signature yield correct predictions when compared to the normalized values with few exceptions.
Genomic signatures for paclitaxel and gemcitabine resistance in breast cancer derived by machine learning. SN Dorman, K Baranova, JHM Knoll, BL Urquhart, G Mariani, ML Carcangiu, PK Rogan. Mol Oncol. 2016 Jan;10(1):85-100. READ PAPER HERE
Predicting Outcomes of Hormone and Chemotherapy in the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) Study by Biochemically-inspired Machine Learning. EJ Mucaki, I Rezaeian, K Baranova, HQ Pham, D Angelov, A Ngom, L Rueda, PK Rogan. F1000Res. 2016 Aug 31;5:2124. READ PAPER HERE
Predicting Response to Platin Chemotherapy Agents with Biochemically-inspired Machine Learning. EJ Mucaki, JZL Zhao, DJ Lizotte, PK Rogan. Signal Transduct Target Ther. 2019. Jan 11;4:1. READ PAPER HERE
Pathway-extended gene expression signatures integrate novel biomarkers that improve predictions of patient responses to kinase inhibitors. AJ Bagchee-Clark, EJ Mucaki, T Whitehead, PK Rogan. MedComm. 2020. doi: 10.1002/mco2.46 READ PAPER HERE
Multigene signatures of responses to chemotherapy derived by biochemically-inspired machine learning. PK Rogan. Mol Genet Metab. 128 (2019): 45-52. READ PAPER HERE