(***) [Machine Learning course at Coursera](coursera.org/learn/machine-learning) by Andrew Ng.
A required course for any people who are interested in machine learning.
(***) [Blog posts by Adam Geitgey](https://medium.com/@ageitgey/machine-learning-is-fun-80ea3ec3c471), his code examples (tensorflow + jupyter) are amazing ([GitHub link](https://github.com/aymericdamien/TensorFlow-Examples)).
Step by step introduction to machine learning using tensorflow.
(*) [Track deep learning related papers on Arxiv](http://www.arxiv-sanity.com/)
(**) [nice blog posts from a stanford graduate Andrej Karpathy](http://karpathy.github.io/).
A few excellent posts to introduce deep learning from a software engineer perspective.
(*) [learn important features from deep learning models](https://arxiv.org/abs/1605.01713)
(**) [Berkeley CS294-112 Deep Reinforcement Learning Sp17](https://www.youtube.com/playlist?list=PLkFD6_40KJIwTmSbCv9OVJB3YaO4sFwkX)
(*) [Deep Learning 2016: The Year in Review](http://www.deeplearningweekly.com/blog/deep-learning-2016-the-year-in-review)
(*) [stanford deep learning in genomics and biomedicine](https://canvas.stanford.edu/courses/51037/files/folder/LectureMaterial)
(*) [google Udacity deep learning course](https://classroom.udacity.com/courses/ud730)
# Code
(*) [Keras Documentation](https://keras.io/)
(*) [deep learning for sequence data DragoNN](http://kundajelab.github.io/dragonn/)
(*) [deep learning for chemical data: deepchem](http://deepchem.io/)
(*) [Open AI generative model post](https://blog.openai.com/generative-models/#contributions)
(*) [introduction-generative-adversarial-networks-code-tensorflow with github code](http://blog.aylien.com/introduction-generative-adversarial-networks-code-tensorflow/)
(**)[tips and tricks training GAN](https://github.com/soumith/ganhacks).
If you have trouble debugging GAN, read this.
(***) [CycleGAN: Software that can generate photos from paintings, turn horses into zebras, perform style transfer, and more.](https://github.com/junyanz/CycleGAN).
Very inspiring application of GAN in domain knowledge translation.
(***) [Deep learning with cats](https://github.com/AlexiaJM/Deep-learning-with-cats).
[Leveraging big data to transform target selection and drug discovery](https://www.ncbi.nlm.nih.gov/pubmed/26659699).
[Harnessing big ‘omics’ data and AI for drug discovery in hepatocellular carcinoma](https://www.nature.com/articles/s41575-019-0240-9).
Data Types include Genomics (G), Epigenomics (E), Transcriptomics (T), Proteomics (P), Metabolomics (M), Phenomics (P), Imaging (I) .
Mainly list the big data repositories
# Disease models
## Clinical models
## Preclinical models
# Perturbagen
## OMICS
(***) [A Library of Phosphoproteomic and Chromatin Signatures for Characterizing Cellular Responses to Drug Perturbations](https://www.biorxiv.org/content/early/2017/09/07/185918).
profiled 90 drugs (in triplicate) in six cell lines using two different proteomic assays
(*) [Orthotopic patient-derived xenografts of paediatric solid tumours](https://www.nature.com/nature/journal/v549/n7670/full/nature23647.html).
sensitivity data of 150 drugs across >20 pediatric cell lines
5,000 pairs of FDA-approved drugs that were tested against the NCI-60
# Single cells
(***) Liver cancer single cells ([Landscape of Infiltrating T Cells in Liver Cancer Revealed by Single-Cell Sequencing](http://www.cell.com/cell/fulltext/S0092-8674(17)30596-2)).
5,063 single T cells isolated from peripheral blood, tumor, and adjacent normal tissues from six hepatocellular carcinoma patients.
(*) The history of the [development of analogs of thalidomide](https://en.wikipedia.org/wiki/Development_of_analogs_of_thalidomide). Thalidomide is one of the most successful repurposed drug.
# Business thoughts
(*) [Escaping Big Pharma’s Pricing With Patent-Free Drugs](https://www.nytimes.com/2017/07/18/opinion/escaping-big-pharmas-pricing-with-patent-free-drugs.html)..
(*) [Cost of Developing a Single Cancer Drug: time 7.3 years, cost $0.6 billion, revenue 1.7 billion](https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/2653012?utm_source=twitter&utm_campaign=content-shareicons&utm_content=article_engagement&utm_medium=social&utm_term=091117#.WbdeR9XhT0M.twitter)
A quick check list of basics in data science (credit to Maverick Lin)
(***) Single cell RNA-Seq analysis [osca](https://osca.bioconductor.org/introduction.html), [seurat](https://satijalab.org/seurat/vignettes.html).
outstanding framework for scRNA-Seq analysis.
# Fundamental papers
**These papers I read at least 10 times, including supplementary materials! All of them are three stars!**
## Field review
(***) [Hallmarks of Cancer: The Next Generation](http://www.cell.com/abstract/S0092-8674%2811%2900127-9), by Robert A. Weinberg.
Fundamental to understand cancer.
(***) [Tumor Metastasis: Molecular Insights and Evolving Paradigms](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3261217/), by Robert A. Weinberg.
Fundamental to understand cancer metastasis.
(***) [Cancer genome landscapes](http://science.sciencemag.org/content/339/6127/1546.long), by Bert Vogelstein.
Fundamental to understand cancer genomics.
(***) [Cancer transcriptome profiling at the juncture of clinical translation](https://www.nature.com/articles/nrg.2017.96), by Arul M. Chinnaiyan.
review on cancer transcriptomics.
(***) [Ewing sarcoma: historical perspectives, current state-of-the-art, and opportunities for targeted therapy in the future.](https://www.ncbi.nlm.nih.gov/pubmed/18525337).
A typical review on the therapeutic discovery of one cancer.
(***) [Opportunities and challenges in phenotypic drug discovery: an industry perspective](https://www.nature.com/nrd/journal/v16/n8/abs/nrd.2017.111.html).
Our drug discovery approach is one type of phenotypic screening.
(***) [Ten Years of Pathway Analysis: Current Approaches and Outstanding Challenges](http://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1002375), by Purvesh Khatri.
A very nice summary of method development in pathway analysis.
(***) [High-performance medicine: the convergence of human and artificial intelligence](https://www.nature.com/articles/s41591-018-0300-7) by Eric Topol.
Current progress and challenges in applying DL into biomedical research.
## Statistical method development
(***) [Significance analysis of microarrays applied to the ionizing radiation response](http://www.pnas.org/content/98/9/5116.full), by Robert Tibshirani.
Development of SAM, a popular method to perform differential expression analysis using microarray data.
(***) [limma: Linear Models for Microarray Data](https://link.springer.com/chapter/10.1007/0-387-29362-0_23).
Development of LIMMA, another popular method to perform differential expression analysis using microarray data.
(***) [Differential expression analysis for sequence count data](https://genomebiology.biomedcentral.com/articles/10.1186/gb-2010-11-10-r106), by Simon Anders.
Development of DEseq, a popular method to perform differential expression analysis using RNA-SEQ data.
(***) [Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles](http://www.pnas.org/content/102/43/15545.long), by Jill P. Mesirov.
Development of GSEA, the most popular gene set enrichment analysis method and the fundamental to understand our drug discovery method.
(***) [Adjusting batch effects in microarray expression data using Empirical Bayes methods](https://academic.oup.com/biostatistics/article/8/1/118/252073/Adjusting-batch-effects-in-microarray-expression).
Development of Combat, a method to correct batch effects.
(***) [Emergence of Scaling in Random Networks](http://science.sciencemag.org/content/286/5439/509.full) by Albert-László Barabási.
Discovery of scale-free networks.
(***) [Pathsim: Meta path-based top-k similarity search in heterogeneous information networks](http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.220.2455) by Jiawei Han.
A typical machine learning approach to mining heterogeneous networks.
(***) [MuSiC: Identifying mutational significance in cancer genomes](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3409272/), by Li Ding.
Development of MuSic, a popular method to identify mutations.
## Informatics method development and application
(***) [The Connectivity Map: using gene-expression signatures to connect small molecules, genes, and disease.](http://science.sciencemag.org/content/313/5795/1929.long) by Justin Lamb.
The first paper to describe our drug discovery approach.
(***) [Discovery and Preclinical Validation of Drug Indications Using Compendia of Public Gene Expression Data](http://stm.sciencemag.org/content/3/96/96ra77) from Atul's lab.
The basic of our drug discovery work, and a great demonstration of writing a computational paper (from method development to experimental validation).
(***) [Relating protein pharmacology by ligand chemistry](http://www.nature.com/nbt/journal/v25/n2/full/nbt1284.html) by Michael J Keiser and Brian K Shoichet.
The development of SEA, a method to predict drug-target interactions, and another great demonstration of writing a computational paper.
(***) [Characterization of drug-induced transcriptional modules: towards drug repositioning and functional understanding](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3658274/) by Peer Bork.
start with data analysis and end with a few biological experiments.
(***) [Cross-Species Regulatory Network Analysis Identifies a Synergistic Interaction between FOXM1 and CENPF that Drives Prostate Cancer Malignancy](http://www.cell.com/cancer-cell/fulltext/S1535-6108(14)00125-1) by Andrea Califano.
start with data analysis and end with a few biological experiments.
(***) [Elucidating compound mechanism of action by network perturbation analysis](http://www.sciencedirect.com/science/article/pii/S0092867415006996) by Andrea Califano.
start with data analysis and end with a few biological experiments.
(***) [Discovery of drug mode of action and drug repositioning from transcriptional responses](http://www.pnas.org/content/107/33/14621.long).
start with data analysis and end with a few biological experiments.
(***) [Imagenet classification with deep convolutional neural networks](http://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf).
Development of convolutional neural networks (CNN), the popular deep learning method.
## Computational analysis
(***) [Drug-target network](https://www.nature.com/nbt/journal/v25/n10/full/nbt1338.html) by Barabási.
Network analysis of drug-target interactions.
(***) [Comprehensive molecular portraits of human breast tumours](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3465532/) from TCGA.
A typical genomic analysis paper from TCGA.
(***) [Mutational landscape and significance across 12 major cancer types.](http://dx.doi.org/10.1038/nature12634), by Li Ding.
A phenomenal paper on pan-cancer genomic analysis.
(***) [Comprehensive Characterization of Molecular Differences in Cancer between Male and Female Patients](http://www.cell.com/cancer-cell/fulltext/S1535-6108(16)30111-8), by Han Liang.
A phenomenal paper on pan-cancer genomic analysis.
(***) [Genetics of rheumatoid arthritis contributes to biology and drug discovery.](https://www.nature.com/nature/journal/v506/n7488/full/nature12873.html) by Robert M. Plenge.
great work using genetics for drug discovery.
(***) [The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity](https://www.nature.com/nature/journal/v483/n7391/full/nature11003.html).
phenomenal work using cell line data to discover biomarkers.
(***) [A comprehensive time-course–based multicohort analysis of sepsis and sterile inflammation reveals a robust diagnostic gene set](http://stm.sciencemag.org/content/7/287/287ra71.short) by Purvesh Khatri.
Phenomenal work using public microarray data to discover biomarkers.
(***) [Prediction of biological targets for compounds using multiple-category Bayesian models trained on chemogenomics databases](http://pubs.acs.org/doi/10.1021/ci060003g) by Jeremy Jenkins.
A typical machine learning paper in cheminformatics.
(***) [Do structurally similar molecules have similar biological activity](https://dx.doi.org/10.1021/jm020155c).
A typical data analysis paper in cheminformatics.
## Deep-learning based drug discovery
(***) [Predicting Drug Response and Synergy Using a Deep Learning Model of Human Cancer Cells](https://pubmed.ncbi.nlm.nih.gov/33096023/) by Ideker.
Develop a model to predict drug activity based on a huge pharmacogenomics dataset, propose novel ways to model cells based on Gene Ontology, and experimentally validate some hits.
(***) [A Deep Learning Approach to Antibiotic Discovery](https://pubmed.ncbi.nlm.nih.gov/32084340/) by Barzilay and Collins.
Develop a model to predict antibiotic activity based on chemical structure, screen millions of compounds and extensively validate one drug candidate.
(***) [Deep reinforcement learning for de novo drug design](http://advances.sciencemag.org/content/4/7/eaap7885) by Tropsha.
Develop a model to generate targeted chemical libraries of novel compounds optimized for either a single desired property or multiple properties.
(***) [Convolutional Networks on Graphs for Learning Molecular Fingerprints](https://arxiv.org/abs/1509.09292) by Ryan Adams.
Designed DL based fingerprints inspired by ECFP
(***) [Automatic Chemical Design Using a Data-Driven Continuous Representation of Molecules](https://pubs.acs.org/doi/full/10.1021/acscentsci.7b00572).
Used Variational autoencoder to encode SMILES and optimize compounds from the latent space.
## Shape our future
(***) [Single-cell RNA-seq highlights intratumoral heterogeneity in primary glioblastoma](http://science.sciencemag.org/content/344/6190/1396).
Application of single cell in a cancer study.
(***) [Single-cell transcriptomics uncovers distinct molecular signatures of stem cells in chronic myeloid leukemia](https://www.nature.com/nm/journal/v23/n6/full/nm.4336.html).
Application of single cell analysis toward personalized cancer therapy.
(***) [Brown Adipogenic Reprogramming Induced by a Small Molecule](http://www.sciencedirect.com/science/article/pii/S2211124716317697) by Sheng Ding.
Using small molecules to control cell development.
(***) [Correlating chemical sensitivity and basal gene expression reveals mechanism of action.](https://www.nature.com/nchembio/journal/v12/n2/full/nchembio.1986.html) from Stuart Schreiber.
Usage of pharmacogenomics data to understand drug mechanisms.
(***) [A Next Generation Connectivity Map: L1000 Platform And The First 1,000,000 Profiles](https://www.biorxiv.org/content/early/2017/05/10/136168) from Todd R. Golub.
LINCS, the dataset we primarily used for drug discovery.
(***) [Integrative clinical genomics of metastatic cancer](http://www.nature.com/nature/journal/v548/n7667/full/nature23306.html).
We have lots of experience working on primary cancer, now it's time to place our interest to metastatic cancer, which the majority of patients die from.
(***) [Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks](https://arxiv.org/pdf/1703.10593.pdf).
Using deep learning GAN to realize domain knowledge translation.
# Outstanding tools and datasets for translational drug discovery
**use liver cancer as an example, can be applied to other cancers, only list outstanding tools/datasets for liver cancer drug discovery**
two review articles from the lab:
[Harnessing big ‘omics’ data and AI for drug discovery in hepatocellular carcinoma](https://www.nature.com/articles/s41575-019-0240-9).
[Leveraging big data to transform target selection and drug discovery](http://www.ncbi.nlm.nih.gov/pubmed/26659699).
Search drugs used in liver cancer clinical trials.
(**) [Cancer Today (Globocan): Data visualization tools that present current national estimates of cancer incidence, mortality, and prevalence](http://gco.iarc.fr/today/home).
create a comprehensive preclinical reference map connecting tumor features with tumor dependencies to accelerate the development of precision treatments.
(***) [GEO](https://www.ncbi.nlm.nih.gov/geo/).
Search functional genomics data for liver cancer, requiring additional computational analysis to create a liver cancer signature.
(*) [Easy Way to Mix Multiple Graphs on The Same Page](http://www.sthda.com/english/articles/24-ggpubr-publication-ready-plots/81-ggplot2-easy-way-to-mix-multiple-graphs-on-the-same-page/).
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.