A planned open-source scoring study. Pilot docking and a 0.25 ns local equilibration run are done. 100 ns runs are planned. No lab validation yet.
Hello,
Tuberculosis (TB) is a serious lung disease. It spreads through the air when an infected person coughs or sneezes.
Even today, more than a million people die from TB every year. Patients have to take strong medications for 6 months or more. Many strains of TB bacteria have become resistant to older drugs (such as isoniazid). So we really need new drugs that can directly stop these becteriya.
My name is Navneet Dwivedi. I am an independent researcher from Singrauli, Madhya Pradesh (India). I have a B.com degree. I live with SLD (Specific Learning Disability), ADHD, autism, and hearing-related learning difficulties.
I am building a tool called "TB-Dynamic-Selectivity v3.0," which is completely free and open-source.
My code is here: https://github.com/navneetdwivedi147-code/TB-Dynamic-Selectivity
TB bacteria need an enzyme called InhA (PDB: 4TZK) to make their cell wall. My tool new drugs that can looks for block this enzyme.
Normal computer docking checks drugs against a static (frozen) snapshot of a protein. But inside a real human body at 310 K, water and heat keep the protein moving all the time.
My tool Checks drugs while the protein is moving. It sees how long a drug stays near the main Tyr158 site. It also compares binding energy (in kcal/mol) between the main target and off-target sites.
This is only a computer model right now, and no lab tests have been done yet. I have already run 310 K GROMACS equilibration tests on my own laptop for my lead candidate MTB-NL4.
Goal:
I want to build a free tool that ranks drug candidates for tuberculosis (TB). The tool checks how a molecule behaves with the protein InhA at human body temperature (310 K). My goal is to get more accurate drug ranking compared to using standard docking scores alone.
How I will do this (if funded):
If I get funded, I will work on this full-time for three month
• Month 1 : Check my docking outputs, verify IC50 numbers from published papers, rebuild the InhA-NAD+ setup, and dock about 50 known TB inhibitors and decoys.
• Month 2 : Rent cloud GPUs, run a speed test, and run 100 ns GROMACS simulations at 310 K for the top candidates, including MTB-NL4.
• Month 3 : Measure contacts with Tyr158 and NAD+ plus buried surface area, tune the scoring weights, get an outside structural biologist to review my setup, and upload all data to Zenodo.
I will know it worked if:
• The 100 ns simulations finish without any errors.
• The reference drugs stay within 3.5 Å of Tyr158 for most frames (I will set the exact percentage before starting the runs.
• The 310 K score ranks known TB inhibitors better than static Vina docking alone.
• An outside structural biologist checks my simulation files and finds no major flaws.
It did not work if:
• I cannot finish the simulations within my budget.
• The 310 K score does not rank the inhibitors better than static docking.
How will this funding be used?
• Living Expenses and Relocation to Lucknow (3 months) – $1,950: $650/month to cover rent, food, and reliable electricity/internet, so I can move to Lucknow next month and work on this project full-time (6 days a week).
• Cloud GPU Compute – $1,200: About 2,100 to 2,400 GPU hours on RunPod or Total Goal: $5,000 | Minimum Goal: $2,500
• Living Expenses and Relocation to Lucknow (3 months) – $1,950: $650/month to cover rent, food, and reliable electricity/internet, so I can move to Lucknow next month and work on this project full-time (6 days a week).
• Cloud GPU Compute – $1,200: About 2,100 to 2,400 GPU hours on RunPod or Vast.ai (around $0.50–$0.55/hour) to run 100 ns GROMACS simulations for 10 to 12 complexes.
• External Structural Biology Review (Conditional) – $650: I will spend this only after a qualified structural biologist finishes reviewing my setup.
• Hardware Storage and RAM Upgrades – $450: Extra RAM, an NVMe SSD, and an external hard drive to safely store raw GROMACS trajectory files.
• Bank Transfer and Currency Conversion Fees – $400: Fees for international wire transfers, platform fees, and INR currency conversion.
• Analysis and Cloud Storage Tools – $350: Cloud storage and analysis tools for three months
(If I only raise the $2,500 minimum, I will reduce the simulations from 10–12 complexes down to 4–6 core complexes (x641, MTB-NL4, top analogues, and key literature controls) and spend the money only on cloud GPU hours, storage, transfer fees, and basic living costs.)
Who is on your team? What's your track record on similar projects?
I am currently working solo as the only developer and researcher, and I plan to hire an outside, aur in Month 3.
My track record so far:
• Built and published the open-source TB-Dynamic-Selectivity v3.0 codebase (code on GitHub (over 20 Python scripts for docking audits, benchmarks, and energy checks)) on GitHub.
• Designed and improved the v3.0 scoring formula across three versions (v1.0 to v3.0) and checked all physical units.
• my new candidates (MTB-NL4 series) and NAD+ using AmberTools (antechamber and GAFF2), and finished GROMACS energy minimization plus NVT and NPT equilibration runs at 310 K on my own laptop.
GPU speed slower than expected: If the speed (ns/day) on the cloud GPU is slower than I expected, I will reduce the number of complexes from 10–12 to 4–6 and share the exact run times.
310 K simulations do not beat static docking: If the 310K MD features do not improve inhibitor ranking compared to Vina, I will still publish the full trajectories, scripts, and the negative results on Zenodo and GitHub, so other researchers do not repeat the same work.
Cannot find an outside reviewer in time: If I cannot find a qualified structural biologist within 3 months, I will not spend the $650 review budget.
How much money have you raised in the last 12 months, and from where?
I have not received any funding yet. I have applied to Emergent Ventures, and I will update this section once I hear back from them.
Comments
No comments yet. Sign in to create one!