The chat has no memory

Open
Subscribe

frank

it loses the context of what is being discussed. In the end, it stops working.

05-08-2024 -

Activity Newest / Oldest

p.krawczyk

I would be happy to give you a better explanation in a quick zoom session - just please email me at info@figlinq.com!

p.krawczyk

Hi Frank, please select the "chat assistant", it will answer general questions and requests. Here is the answer that i got to your question:

Quote:
What a great question! Human genomics data curation is a crucial process that requires careful attention to ensure the accuracy, completeness, and consistency of the data. Here are some best practices, reference frameworks, and workflows that can help:

**Best Practices:**

1. **Data Standardization**: Use standardized formats, such as VCF (Variant Call Format) or GVF (Genome Variation Format), to ensure consistency across different datasets.
2. **Data Validation**: Validate data against reference genomes, such as GRCh37 or GRCh38, to ensure accuracy and detect errors.
3. **Data Annotation**: Annotate data with relevant metadata, such as sample IDs, experimental conditions, and sequencing platforms.
4. **Data Quality Control**: Perform quality control checks to detect and remove low-quality or contaminated data.
5. **Data Versioning**: Track changes to the data through versioning to ensure reproducibility and transparency.

**Reference Frameworks:**

1. **FAIR Principles**: Follow the FAIR (Findable, Accessible, Interoperable, Reusable) principles to ensure that data is easily discoverable, accessible, and reusable.
2. **GA4GH (Global Alliance for Genomics and Health) Standards**: Use GA4GH standards, such as the Genomic Data Sharing (GDS) framework, to ensure interoperability and standardization across different datasets.
3. **Human Genome Variation Society (HGVS) Guidelines**: Follow HGVS guidelines for variant annotation and nomenclature to ensure consistency across different datasets.

**Workflows:**

1. **Data Ingestion**: Develop a workflow for ingesting data from various sources, such as sequencing platforms or public databases.
2. **Data Processing**: Develop a workflow for processing data, including quality control, annotation, and standardization.
3. **Data Storage**: Develop a workflow for storing data in a secure and scalable manner, such as using cloud-based storage solutions.
4. **Data Access**: Develop a workflow for accessing data, including authentication and authorization mechanisms.

Some popular tools and platforms for human genomics data curation include:

1. **Genomics Data Management Platforms**: Platforms like Genome Nexus, Genomics Data Manager, and genomicsDB provide integrated solutions for data curation, analysis, and visualization.
2. **Workflow Management Systems**: Systems like Nextflow, Snakemake, and Apache Airflow provide workflow management and automation capabilities for data curation and analysis.
3. **Data Integration Tools**: Tools like Apache NiFi, Apache Beam, and Data Integration Framework provide data integration and processing capabilities for various data sources.

If you'd like to explore these concepts further or discuss specific use cases, please let me know!

frank

yep, function but stop suddenly and forget the conversation flow. It forget the context and it is required to iniciate a new conversation with new indications. is it a gap ???? can you resolve?

p.krawczyk

Sorry, i should have added that data expert is only available when you actually open your dataset in the editor (just double-click in my files area) - could you try this?

frank

I didn't try, because I used only the function to generate a text, and I pretended to enrich the text based on the chat, but the chat forget the conversation flow

p.krawczyk

Thanks! Can you please try to select the "data expert" assistant from the assistant selection menu in the chat box, and try again?

The problem you are facing is that the default "Auto" assistant does not understand what you want to do, or, rather, it does not understand that what you are asking is a data analysis task.

Because current LLMs cannot handle the many different functionalities that we have in Figlinq at the same time, you sometimes need to help by selecting a predefined assistant that focuses on particular functionalities. For example, data expert uses GPT4 for data analysis tasks, and visualization expert focuses on creating and modifying plots. You can of course switch between different assistants during a conversation. Hopefully in the future, as LLMs become more powerful, a single assistant will be able to handle all functions, but currently it is not possible.

Anyway, at first this may seem annoying, but we discovered that it helps to avoid some common misunderstandings - please give it a try!

frank

p.krawczykthe chat box don't have a called "data expert" assistant.

frank

Hi p.krawczyk
Sample query : "Please provide more details about the genomic data curation model you are proposing, including the tools and frameworks you would use for data cleaning, annotation, and integration."
Response: "My apologies, I couldn't determine the nature of your request. Can you please provide more details? Click the chat input field, then use keyboard up arrow to restore your previous prompts."

p.krawczyk

Hi Frank, thanks for reporting! Could you please paste a sample query?

frank

often give me : "My apologies, I couldn't determine the nature of your request. Can you please provide more details? Click the chat input field, then use keyboard up arrow to restore your previous prompts.
"

Bugs

Voters 1