This project is dedicated to developing a set of automated tools and processes for extracting useful neuronal experimental data from datasets provided by the Allen Brain Institute, and converting these data into parameters that can be used for neuronal modeling on the NetPyNE platform, for comparison with cortical regions already validated by NetPyNE. The project flow is divided into the following steps: Data Collection: Automatically extracts data from the Allen Cell Type Database for specific cell types in the visual cortex, including electrophysiological properties, morphological characteristics, and gene expression patterns. The Allen Institute provides a powerful API that allows users to programmatically access and retrieve data. First through data cleaning, and then through data crawling to obtain data that is difficult to access in the API. In addition, domain-specific language models will be fine-tuned or trained using Google's Gemini model for subsequent data acquisition. Parameter Conversion: Develop tools(scripts) or design large language model prompts to automatically convert extracted data into a parameter format understood by the NetPyNE, such as .hoc or .swc format. Identify how to set reasonable default values for missing data. Model construction: Use the converted parameters to construct neural network models of relevant brain regions in NetPyNE. This includes defining neuron types, neuron connectivity, and synaptic interactions between neurons. Run Simulation: Run a simulation of the model in NetPyNE to simulate the brain's response to a specific input stimulus. Compare Results: Compare the simulation results with actual biological experimental data to verify the accuracy of the model. Intended set of deliverables: Automated Data Collection Tool. Parameter Conversion Script. Simulation Run Framework. Result Validation Report. Documentation and User Guide.

• Implement a Project edit button, file and project rename functionality, fix edit project popup , allow editing list of notebooks , and add images to each type of notebook. • Implement support for Jupyter notebooks. Users should be able to upload, view, and edit Jupyter notebooks on the platform. Use Jupyter Lite, or nbviewer, or nbconvert to render notebooks in HTML format. • Allow users to fork existing projects. Forking should create a new project that is linked to the original, allowing users to experiment with changes without affecting the original project. • Improve the project creation and editing process. Add validations for project names to ensure uniqueness. Enhance the project edit functionality to prefill forms with existing information and update project names and routes when edited. • Implement a user permission system to differentiate between project members and followers. Users can join projects as members or follow them with the project owner's permission. • Set up a continuous integration system for automated Docker container generation. • Migrate the project's blob storage to Jetstream. Ensure a seamless transition with minimal downtime and data loss. Set up automated, regular database backups. • Bug Fixes: Address existing bugs in the platform, such as issues with the registration code, disappearing navbar account menu, and join button behavior, to improve user experience and platform stability. • Implement an organization system where users can create and manage organizations. Organizations can have multiple projects and users associated with them. Define roles such as admin, members, and followers within organizations to manage permissions and access levels. • Implement comments in projects and logs along with a reply-to field, vote field to up/down voting, and a ‘flag’ field for flagging.

To participate in the Neurobagel query federation, datasets must conform to Neurobagel’s data model, so annotating the datasets is necessary to harmonize them for query federation. The project aims to reduce the human effort to manually annotate individual data elements by automating the current annotation tool provided by Neurobagel using Large Language Models (LLMs). The already existing annotation tool will be integrated with an LLM-based assistant which will categorize and annotate each data element and it will be followed by human verification. The project includes automating the tool using LLMs, then integrating the tool into the existing webpage and making changes in the UI accordingly.

My proposal is to implement and expand the social aspects of the project by setting up comments, sharing, following of projects/users, and messaging between users. To do this I will create a messaging interface working within the design language already implemented throughout the site, and incorporate a method to encrypt and store messages on the connected database.

With the advent of machine learning, statistical learning and modelling becoming more prevalent in today’s world, we can estimate values for an unknown data given a trained model i.e. finding value of data(D) given parameter (Θ) .Now what about the reverse? What if we want to find optimal parameter value, given a particular set of dataset. Why is this even required? The very simple answer to this can be found in every modelling and computation textbook. We cannot necessarily measure the other parameters. Let us take an example: There are several parameters while defining a neuron whose measurement is difficult to make, for example the axial resistance, the capacitance of the soma etc. Thus to achieve this process we make use of 2 tools, the first is the neuron library which is used for designing your neuron model and setting your parameters ,secondly is a statistical measurement tool used for the parameter estimation .This is done by checking for the voltage value generated by the neuron model and what might the parameter values be for that particular value. The reason to utilize Bayesian inference for parameter estimation is due to the specification of the prior data ,which will help us to increase the accuracy of our model which predicts approximate parameter values. Now the user could specify the prior data on the basis of any previous study or research material. Neuroptimus is an open source tool created for the same purpose. It is an interactive tool where the user can drop a dataset file. Train it and use it for parameter estimation . I have been tasked to utilize the prior ,the noise (for the calculation of likelihood),and the evidence to find the posterior distribution of the parameter I want to estimate. This is the main gist of the project, but there are various hurdles such as what if it is difficult to find the likelihood ,type of noise, computational costs for multiple parameters etc. These are some other things I believe I will be working on.

The proposal focuses on prospective ideas that can be employed for enhancement of the existing INCF Impact Visualization Portal. It goes into depth, explaining every idea's importance and possible way of implementation. Additionally, it proposes novel ideas, as well as ideas for upgrading the existing features.

This proposal aims to improve the capability of markdown exporter package inorder to cover multicompartment model, and also to support generation of p.d.f, h.t.m.l , latex file types not only this but also to follow a standard template generation using jinja2 template generation tool , Inorder to streamline the process for user to select the output according to their requirements based on their template selection.

The project aims to refactor the Brain's unit system into a standalone package. Additionally, it aims to incorporate additional functionalities from other unit packages such as astropi (for type annotations), pint (for integration with matplotlib’s unit system), Quantiphy (for string formatting), and pint-pandas (for integration with pandas). To achieve the desired outcome, I propose structuring the standalone package into multiple submodules, each well-documented with READMEs and examples. Moreover, the package will include new units, constants, and potentially implementation of tests for other projects. This approach will enable other organizations and developers to easily utilize the resulting Brain unit package in their projects. The integration into other projects will be seamless and effortless due to the well-structured documentation and package stability. This will ensure widespread adoption of the Brain unit package in scientific research projects, thereby making a significant impact on the computational and development communities.

The ‘Open Source Community Sustainability’ project is about simulating open-source community interactions to encourage conditions leading to sustainable practice. I propose to do this by creating a framework with an agent-based model powered by Large Language Models (LLMs). This project has been worked on in OREL for the past 2 years, and a variety of models and tools have been created towards it, utilizing reinforcement learning, active inference and attraction-repulsion based approaches to build several agent-based models. Thus, this framework will represent open-source communities as a system composed of interconnected components (agents), mapping the intricacies of community engagement and collaboration as well as focusing on emergent properties and overall system behaviour. This project is important to the open-source ecosystem because as the demand for open-source software continues to grow, so do the challenges associated with community management, collaboration, and sustainability. Thus, providing a framework to understand how individuals collectively contribute to community growth and success can help derive insights into effective community management, collaboration strategies, and sustainability factors.

The Active Segmentation Platform for ImageJ (ASP/IJ) aims to provide a general-purpose workbench that allows biologists and other domain experts to access state-of-the-art techniques in machine learning to achieve excellent image segmentation and classification. This project focuses on enhancing the functionality and user experience of ASP/IJ by updating its user interfaces and developing a new visualization and reporting panel using the Weka (Waikato Environment for Knowledge Analysis) library. This panel will enable users to easily utilize Weka's advanced visualization and analysis tools, thereby facilitating image segmentation and classification.