A data-driven approach to the automated mapping of functional brain topographies across species, bioRxiv, 2018-09-09
AbstractBehavioral neuroscience has made great strides in developing animal models of human behavior and psychiatric disorders. Animal models allow for the formulation of hypotheses regarding the mechanisms underlying psychiatric disorders, and the opportunity to test these hypotheses using procedures that are too invasive for human participants. However, recent scientific reviews have highlighted the low success rate of translating results from animal models into clinical interventions in humans. A potential roadblock is that bidirectional functional mappings between the human and rodent brain are incomplete. To narrow this gap, we created a framework, Neurobabel, for performing large-scale automated synthesis of human neuroimaging data and behavioral neuroscience data. By leveraging the semantics of how researchers within each field describe their studies, this framework enables region to region mapping of brain regions across species, as well as cross-species mapping of psychological functions. As a proof of concept, we utilize the framework to create a functional cross-species mapping between the amygdala and hippocampus for fear-related and spatial memories, respectively. We then proceed to address two open questions in the field (1) Do rodents have a dorsolateral prefrontal cortex? (2) Which human brain region corresponds to the rodent prelimbic cortex?
biorxiv neuroscience 0-100-users 2018A data-driven approach to the automated study of cross-species homologies, bioRxiv, 2018-09-09
AbstractBehavioral neuroscience has made great strides in developing animal models of human behavior and psychiatric disorders. Animal models allow for the formulation of hypotheses regarding the mechanisms underlying psychiatric disorders, and the opportunity to test these hypotheses using procedures that are too invasive for human participants. However, recent scientific reviews have highlighted the low success rate of translating results from animal models into clinical interventions in humans. A potential roadblock is that bidirectional functional mappings between the human and rodent brain are incomplete. To narrow this gap, we created a framework, Neurobabel, for performing large-scale automated synthesis of human neuroimaging data and behavioral neuroscience data. By leveraging the semantics of how researchers within each field describe their studies, this framework enables region to region mapping of brain regions across species, as well as cross-species mapping of psychological functions. As a proof of concept, we utilize the framework to create a functional cross-species mapping between the amygdala and hippocampus for fear-related and spatial memories, respectively. We then proceed to address two open questions in the field (1) Do rodents have a dorsolateral prefrontal cortex? (2) Which human brain region corresponds to the rodent prelimbic cortex?
biorxiv neuroscience 0-100-users 2018Resource Scalable whole genome sequencing of 40,000 single cells identifies stochastic aneuploidies, genome replication states and clonal repertoires, bioRxiv, 2018-09-07
SummaryEssential features of cancer tissue cellular heterogeneity such as negatively selected genome topologies, sub-clonal mutation patterns and genome replication states can only effectively be studied by sequencing single-cell genomes at scale and high fidelity. Using an amplification-free single-cell genome sequencing approach implemented on commodity hardware (DLP+) coupled with a cloud-based computational platform, we define a resource of 40,000 single-cell genomes characterized by their genome states, across a wide range of tissue types and conditions. We show that shallow sequencing across thousands of genomes permits reconstruction of clonal genomes to single nucleotide resolution through aggregation analysis of cells sharing higher order genome structure. From large-scale population analysis over thousands of cells, we identify rare cells exhibiting mitotic mis-segregation of whole chromosomes. We observe that tissue derived scWGS libraries exhibit lower rates of whole chromosome anueploidy than cell lines, and loss of p53 results in a shift in event type, but not overall prevalence in breast epithelium. Finally, we demonstrate that the replication states of genomes can be identified, allowing the number and proportion of replicating cells, as well as the chromosomal pattern of replication to be unambiguously identified in single-cell genome sequencing experiments. The combined annotated resource and approach provide a re-implementable large scale platform for studying lineages and tissue heterogeneity.
biorxiv genomics 100-200-users 2018The geometry of abstraction in hippocampus and pre-frontal cortex, bioRxiv, 2018-09-07
The curse of dimensionality plagues models of reinforcement learning and decision-making. The process of abstraction solves this by constructing abstract variables describing features shared by different specific instances, reducing dimensionality and enabling generalization in novel situations. Here we characterized neural representations in monkeys performing a task where a hidden variable described the temporal statistics of stimulus-response-outcome mappings. Abstraction was defined operationally using the generalization performance of neural decoders across task conditions not used for training. This type of generalization requires a particular geometric format of neural representations. Neural ensembles in dorsolateral pre-frontal cortex, anterior cingulate cortex and hippocampus, and in simulated neural networks, simultaneously represented multiple hidden and explicit variables in a format reflecting abstraction. Task events engaging cognitive operations modulated this format. These findings elucidate how the brain and artificial systems represent abstract variables, variables critical for generalization that in turn confers cognitive flexibility.
biorxiv neuroscience 100-200-users 2018Brain-Score Which Artificial Neural Network for Object Recognition is most Brain-Like?, bioRxiv, 2018-09-05
The internal representations of early deep artificial neural networks (ANNs) were found to be remarkably similar to the internal neural representations measured experimentally in the primate brain. Here we ask, as deep ANNs have continued to evolve, are they becoming more or less brain-like? ANNs that are most functionally similar to the brain will contain mechanisms that are most like those used by the brain. We therefore developed Brain-Score – a composite of multiple neural and behavioral benchmarks that score any ANN on how similar it is to the brain’s mechanisms for core object recognition – and we deployed it to evaluate a wide range of state-of-the-art deep ANNs. Using this scoring system, we here report that (1) DenseNet-169, CORnet-S and ResNet-101 are the most brain-like ANNs. (2) There remains considerable variability in neural and behavioral responses that is not predicted by any ANN, suggesting that no ANN model has yet captured all the relevant mechanisms. (3) Extending prior work, we found that gains in ANN ImageNet performance led to gains on Brain-Score. However, correlation weakened at ≥ 70% top-1 ImageNet performance, suggesting that additional guidance from neuroscience is needed to make further advances in capturing brain mechanisms. (4) We uncovered smaller (i.e. less complex) ANNs that are more brain-like than many of the best-performing ImageNet models, which suggests the opportunity to simplify ANNs to better understand the ventral stream. The scoring system used here is far from complete. However, we propose that evaluating and tracking model-benchmark correspondences through a Brain-Score that is regularly updated with new brain data is an exciting opportunity experimental benchmarks can be used to guide machine network evolution, and machine networks are mechanistic hypotheses of the brain’s network and thus drive next experiments. To facilitate both of these, we release <jatsext-link xmlnsxlink=httpwww.w3.org1999xlink ext-link-type=uri xlinkhref=httpBrain-Score.org>Brain-Score.org<jatsext-link> a platform that hosts the neural and behavioral benchmarks, where ANNs for visual processing can be submitted to receive a Brain-Score and their rank relative to other models, and where new experimental data can be naturally incorporated.
biorxiv neuroscience 200-500-users 2018Integrated computational and experimental identification of p53, KRAS and VHL mutant selection associated with CRISPR-Cas9 editing, bioRxiv, 2018-09-04
AbstractRecent studies have reported that CRISPR-Cas9 gene editing induces a p53-dependent DNA damage response in primary cells, which may select for cells with oncogenic p53 mutations11,12. It is unclear whether these CRISPR-induced changes are applicable to different cell types, and whether CRISPR gene editing may select for other oncogenic mutations. Addressing these questions, we analyzed genome-wide CRISPR and RNAi screens to systematically chart the mutation selection potential of CRISPR knockouts across the whole exome. Our analysis suggests that CRISPR gene editing can select for mutants of KRAS and VHL, at a level comparable to that reported for p53. These predictions were further validated in a genome-wide manner by analyzing independent CRISPR screens and patients’ tumor data. Finally, we performed a new set of pooled and arrayed CRISPR screens to evaluate the competition between CRISPR-edited isogenic p53 WT and mutant cell lines, which further validated our predictions. In summary, our study systematically charts and points to the potential selection of specific cancer driver mutations during CRISPR-Cas9 gene editing.
biorxiv cancer-biology 200-500-users 2018