• Testing an open-source scoring tool (TB-Dynamic-Selectivity v3.0) with 100 ns GROMACS runs to rank tuberculosis InhA inhibitors at body temperature.
My name is Navneet Dwivedi. I live in Singrauli, Madhya Pradesh, India. I have a B.Com degree and I also live with SLD, ADHD, Autism, and an auditory learning disability. I am building a free tool called TB-Dynamic-Selectivity v3.0 on GitHub (https://github.com/navneetdwivedi147-code/TB-Dynamic-Selectivity). It helps rank drug candidates against tuberculosis by targeting the InhA enzyme (PDB: 4TZK). Normal docking tools test molecules on a frozen protein picture. In a real human body at 310 K, heat and water make the protein move all the time. My scoring formula checks how well a molecule stays near Tyr158 (within 3.5 A) during this motion and compares target binding against off-target binding in kcal/mol. Right now this is only a computer model. No lab tests have been done yet.
With this grant, I will shift to Lucknow next month and work six days a week for three months. In month 1, I will check my old docking files, fix small notes in my GitHub README, verify IC50 numbers from published papers, rebuild proper holo-InhA setups with NAD+, and dock around 50 TB inhibitors and decoys using AutoDock Vina 1.2.7. In month 2, I will rent cloud GPUs on RunPod or Vast.ai, run a speed test on a ~137,000-atom holo-InhA setup, and run 100 ns GROMACS simulations at 310 K for top candidates like MTB-NL4. In month 3, I will calculate energy, Tyr158/NAD+ contacts, and buried surface area, tune the v3.0 weights with grouped cross-validation, pay an outside structural biologist to review my files, and upload all data to Zenodo.
Total goal is $5,000 (minimum $2,500). For $5,000: $1,950 for 3 months of living costs and rent in Lucknow ($650/month), $1,200 for cloud GPU rent (~2,100 to 2,400 hours at $0.50-$0.55/hr), $450 for extra RAM, SSD, and backup hard drive, $350 for 3 months of cloud tools, $650 reserved to pay an outside structural biologist (spent only if I find a confirmed reviewer), and $400 for bank wire and currency conversion fees. If only $2,500 is raised, I will work from home and spend it on GPU hours, storage, outside review, and transfer fees.
I am working alone as an independent researcher. I set the physics rules and verification checks myself, and I use AI coding assistants to write Python scripts. Earlier versions had bugs that I caught using assert checks in full_pipeline_v3_audit.py (commit 5d8531a) and fixed. All 4 math sanity checks pass, my proxy benchmarks are public on GitHub (commit b227989), and my candidate MTB-NL4 has 0 PAINS alerts with a static Vina 1.2.7 score of -8.56 kcal/mol.
$0 raised so far. I also submitted applications to Emergent Ventures ($5,000) and Experiment.com ($5,000), both under review.
Comments
No comments yet. Sign in to create one!