LSP Data and Methods
LSP investigators perform many relatively conventional molecular cell biology experiment in the process of hypothesis testing, but we concentrate on highly multiplexed assays able to collect many pieces of data from cell extracts, cultured cells, and whole tissues using proteomics, live and fixed-cell imaging, transcript profiling and cytokine profiling. In many cases we develop new methods to collect and analyze these data. The most valuable methods are described in an LSP “microsite” along with associated datasets.
Tissue Cyclic Immunofluorescence (t-CyCIF)
The t-CyCIF method builds on earlier methods involving repeated staining and imaging to generate multiplexed images of formalin-fixed, paraffin-embedded (FFPE) samples using an iterative process (a cycle) in which conventional low-plex fluorescence images are repeatedly collected from the same sample and then assembled into a high dimensional representation. Highly multiplexed images of intact tumor architecture can be used to quantify signal transduction cascades, measure the levels of tumor antigens and determine immunophenotypes using immune cell lineage markers. t-CyCIF is a powerful tool to study drug resistance of immunotherapy in different patients.
Publication: Lin J-R, Izar B, Wang S, Yapp C, Mei S, Shah PM, Santagata S, Sorger PK. Highly multiplexed immunofluorescence imaging of human tissues and tumors using t-CyCIF and conventional optical microscopes. Elife. 2018 Jul 11;7. PMCID: PMC6075866.
Reliable high-throughput imaging of cells grown in multi-plate wells is complicated by the loss of cells during staining and wash steps. The dye drop method uses a set of incrementally more dense solutions to prevent cell loss. Dye Drop software consists of Python tools for determining the viability and cell cycle states of cells before and after drug treatment.
Publication: Hafner M, Mills CE, Gerosa L, Chung M, Niepel M, Sorger PK. DeepDyeDrop: an image-based approach to quantify the phenotypic response of cancer cells to therapeutics (Conference Presentation). In: Evans CL, Chan KF, editors. Visualizing and Quantifying Drug Distribution in Tissue II. San Francisco, United States: SPIE; 2018. p. 13. Available from: https://spiedigitallibrary.org/conference-proceedings-of-spie/10475/2293204/DeepDyeDrop–an-image-based-approach-to-quantify-the-phenotypic/10.1117/12.2293204.full
The Growth Rate inhibition (GR) Calculator (grcalculator.org) is an open source set of Python, R and online tools for quantifying the responses of cancer cells to drugs in a manner that corrects for the confounding effects of variable cell proliferation rates. Response metrics computed from GR data include GR50 and GRmax and are direct analogs of familiar IC50 and Emax response measures.
Publication: Hafner M, Niepel M, Chung M, Sorger PK. Growth rate inhibition metrics correct for confounders in measuring sensitivity to cancer drugs. Nat Methods. 2016 Jun 1;13(6):521–527. PMCID: PMC4887336.
Hafner M, Niepel M, Sorger PK. Alternative drug sensitivity metrics improve preclinical cancer pharmacogenomics. Nat Biotechnol. 2017 Jun 7;35(6):500–502. PMCID: PMC5668135.
HMS LINCS Database
The HMS LINCS Database is a public resource for browsing, searching, and downloading all fully annotated datasets generated through the HMS LINCS Center along with experimental and analytical methods. The database contains >350 datasets from imaging, transcriptomic, and proteomic experiments as well as in vitro small molecule affinity assays.
Ludwig Tumor Atlas
The Ludwig Tumor Atlas project brings together investigators from across the U.S. and Europe to investigate the basis of intrinsic and acquired resistance to anti-cancer drugs and to develop therapies for overcoming resistance. Spatial atlases of cancers that reveal key features of the transformed state in primary and metastatic settings and identify recurrent spatial features predictive of response to therapy.
ProteinNet is a standardized data set for machine learning of protein structure. It provides protein sequences, structures (secondary and tertiary), multiple sequence alignments (MSAs), position-specific scoring matrices (PSSMs), and standardized training / validation / test splits. ProteinNet builds on the biennial CASP assessments, which carry out blind predictions of recently solved but publicly unavailable protein structures, to provide test sets that push the frontiers of computational methodology. It is organized as a series of data sets, spanning CASP 7 through 12 (covering a ten-year period), to provide a range of data set sizes that enable the assessment of new methods in relatively data-poor and data-rich regimes.
Publication: AlQuraishi M. ProteinNet: a standardized data set for machine learning of protein structure. BMC Bioinformatics. 2019 Jun 11;20(1):311. PMCID: PMC6560865.
DRIAD (Drug Repurposing In Alzheimer’s Disease) is a machine learning framework for identifying potential associations between Alzheimer’s Disease pathology and individual genes. DRIAD was applied to data from perturbation experiments performed in the LSP using differentiated human neural cell cultures and 80 FDA-approved and clinically tested drugs, to nominate existing drugs that could be potentially be repurposed for Alzheimer’s treatment.
Publication: Publications: Rodriguez S, Hug C, Todorov P, Moret N, Boswell SA, Evans K, Zhou G, Johnson NT, Hyman BT, Sorger PK, Albers MW, Sokolov A. Machine learning identifies candidates for drug repurposing in Alzheimer’s disease. Nat Commun. 2021 Feb 15;12(1):1033. PMCID: PMC7884393.
DUB (Deubiquitinating enzymes) portal provides the integration and analysis of publicly available resources as well as newly collected transcriptomic data following DUB knockout or inhibition to facilitate the exploration of DUB function in oncology.