Abstract
Despite great advances, molecular cancer pathology is often limited to the use of a small number of biomarkers rather than the whole transcriptome, partly due to computational challenges. Here, we introduce a novel architecture of Deep Neural Networks (DNNs) that is capable of simultaneous inference of various properties of biological samples, through multi-task and transfer learning. It encodes the whole transcription profile into a strikingly low-dimensional latent vector of size 8, and then recovers mRNA and miRNA expression profiles, tissue and disease type from this vector. This latent space is significantly better than the original gene expression profiles for discriminating samples based on their tissue and disease. We employed this architecture on mRNA transcription profiles of 10750 clinical samples from 34 classes (one healthy and 33 different types of cancer) from 27 tissues. Our method significantly outperforms prior works and classical machine learning approaches in predicting tissue-of-origin, normal or disease state and cancer type of each sample. For tissues with more than one type of cancer, it reaches 99.4% accuracy in identifying the correct cancer subtype. We also show this system is very robust against noise and missing values. Collectively, our results highlight applications of artificial intelligence in molecular cancer pathology and oncological research. DeePathology is freely available at https://github.com/SharifBioinf/DeePathology.
Highlights
Despite great advances, molecular cancer pathology is often limited to the use of a small number of biomarkers rather than the whole transcriptome, partly due to computational challenges
Each of our Deep Neural Networks (DNNs) consists of two parts that are serially connected: the encoder part, that learns to convert a given mRNA expression profile of a clinical sample at the input layer to a latent representation, which we call Cell Identity Code (CIC), and the decoder part that infers multiple outputs from the CIC
While the CIC is a simple vector of numerical values in the Contractive Autoencoder (CAE), the Variational autoencoder (VAE) encode the input into two equal-sized vectors, which represent means and standard deviations of multiple Gaussian distributions
Summary
Molecular cancer pathology is often limited to the use of a small number of biomarkers rather than the whole transcriptome, partly due to computational challenges. The challenge is how to train the network in order to learn useful latent representations of the mRNA EPs. To address this challenge, we designed the decoder part of each network to simultaneously learn four different classification and regression problems using the CIC, in a multi-task learning scheme: (i) reproducing mRNA EP as one of the outputs that is as close to the original mRNA EP in the input as possible, (ii) predicting a miRNA expression profile (miRNA EP) that is as close as possible to the experimentally measured miRNA EP of the same sample, (iii) predicting the sample tissue of origin, among 27 different tissues, and (iv) predicting the sample disease state, which can be either normal or one of 33 different cancer types.
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
