This is a general-purpose bioinformatics library with the goal of having excellent
- documentation,
- reproducibility, and
- extensibility
with less debuging. The choice of Haskell is to make it possible for researchers to expertly define workflows and pipelines without requiring expertise in software engineering.
- UMAP
- Tools should be compression agnostic such that the tool receives a feed of data in an uncompressed format and outputs data in an uncompressed format such that any arbitrary decompressor and compressor can be composed with the tool.
I think that maybe this article might be helpful. It suggests using Cancer-associated fibroblasts (file located here.
Started constructing types and began to test a parser implementation using Tsoding's JSON parser YouTube video as reference. The corresponding repo: https://github.com/tsoding/haskell-json
I am recruiting some help to work on this project, and I need to state some clear goals and objectives.
We will want to increase the priority of the AI component in light of the new scholarship/grant opportunity.
What would be really helpful would be a dimensionality-reduction tool to show distances between sequences and molecules, and I think it would make a lot of sense to do point-by-point similarity where the similarity at a point falls off exponentially with distance from the point, which then is computed to the point score. The idea is that if two sequences are separated by a repeat or some slippage or duplication within a region for whatever reason that this will not completely change its similarity score. It seems like the comparison must be done itteratively with mers all the way up to the length of the molecule. Then, the UMAP process should be done because ultimately two sequences can each be similar to a third sequence while having no similarity between themselves.
Starting a new parser specifically for ATCG documents. The goal is to use minimal datatypes for the base so that it can be used for streamlined ML applications without much extra work especially that on purpose-built hardware.