Heckman, C. — Page 2 | Search bioRxiv

bioRxiv2019

Immunogenomic landscape of hematological malignancies

Understanding factors that shape the immune landscape across hematological malignancies is essential for immunotherapy development. Here, we integrated over 8,000 transcriptomes and over 1,000 samples with multilevel genomic data of hematological cancers to investigate how immunological features are linked to cancer subtypes, genetic and epigenetic alterations, and patient survival. Infiltration of cytotoxic immune cells was associated with distinct microenvironmental responses and driver alterations in different cancers, such as TP53 in acute myeloid leukemia and DTX1 in diffuse large B cell lymphoma. Epigenetic modification of CIITA regulating antigen presentation, cancer type-specific immune checkpoints such as VISTA in myeloid malignancies, and variation in cancer antigen expression further contributed to immune heterogeneity. Prognostic models highlighted the significance of immunological properties in predicting survival. Our study represents the most comprehensive effort to date to link immunology with cancer subtypes and genomics in hematological malignancies, providing a resource to guide future studies and immunotherapy development.

immunology↗

bioRxiv2018

Improving the consistency of functional genomics screens using molecular features - a multi-omics, pan-cancer study

Probing the genetic dependencies of cancer cells helps understand the tumor biology and identify potential drug targets. RNAi-based shRNA and CRISPR/Cas9-based sgRNA have been commonly utilized in functional genetic screens to identify essential genes affecting growth rates in cancer cell lines. However, questions remain whether the gene essentiality profiles determined using these two technologies are comparable. In the present study, we collected 42 cell lines representing a variety of 10 tissue types, which had been screened both by shRNA and CRISPR techniques. We observed poor consistency of the essentiality scores between the two screens for the majority of the cell lines. The consistency did not improve after correcting the off-target effects in the shRNA screening, suggesting a minimal impact of off-target effects. We considered a linear regression model where the shRNA essentiality score is the predictor and the CRISPR essentiality score is the response variable. We showed that by including molecular features such as mutation, gene expression and copy number variation as covariates, the predictability of the regression model greatly improved, suggesting that molecular features may provide critical information in explaining the discrepancy between the shRNA and CRISPR-based essentiality scores. We provided a Combined Essentiality Score (CES) based on the model prediction and showed that the CES greatly improved the consensus of common essential genes. Furthermore, the CES also identified novel essential genes that are specific to individual cell types. Taken together, we provided a systematic approach to define a more accurate gene essentiality profile by integrating functional screen data and molecular profiles.

systems biology↗