My top 5 picks

Full list

A. Gut microbiome and drug metabolism (trendy, low cost)

1. β-glucuronidase inhibition assay. Screen plant extracts or agro-waste (pomegranate peel, onion peel, mango kernel) with the p-nitrophenyl-glucuronide assay. Use a bacterial or E. coli enzyme preparation. Add docking against the enzyme and, if possible, a rat or cell toxicity check.
2. Postbiotics for CRC. Test cell-free supernatants of Lactobacillus isolated from local fermented foods (curd, kanji, idli batter) on HT-29 or HCT116. Measure viability, apoptosis, and synergy with 5-FU. Using regional food sources adds novelty.
3. Bacterial metabolism of fluoropyrimidines. Incubate capecitabine or 5-FU with gut bacterial cultures and measure drug depletion by UV or HPLC. The question is whether the microbiome changes how much drug is available.
4. Butyrate as a co-therapy. Test butyrate with 5-FU or oxaliplatin. Butyrate acts as an HDAC inhibitor and behaves differently in normal vs cancer cells (the Warburg-related effect).

B. Drug repurposing (cheap drugs, strong story)

1. Anthelmintic or antifungal plus chemotherapy. Candidates include mebendazole, niclosamide, nitazoxanide, itraconazole and ivermectin. Use Chou-Talalay synergy analysis (CompuSyn is free).
2. Disulfiram plus copper plus 5-FU or oxaliplatin. It induces ROS and cuproptosis-like stress, and the drugs are very cheap.
3. SGLT2 inhibitors, statins, propranolol or sertraline in CRC cells. Read recent literature first to find a gap.
4. Melatonin, vitamin D or metformin with chemotherapy, with Wnt/β-catenin and EMT markers as readouts.

C. Mechanism-based (current topics)

1. Ferroptosis induction. Test a natural compound or repurposed drug. Assays are cheap colorimetric ones: MDA, GSH, ROS, intracellular iron, plus rescue with ferrostatin-1. Ferroptosis is an active CRC area. For example, acevaltrate was reported as a ferroptosis inducer in CRC (Signal Transduct Target Ther 2025).
2. Colonosphere or cancer stem cell inhibition. Measure sphere formation, ALDH activity and CD44 expression. Stem cells are a main cause of relapse.

D. In vivo and alternative models (low ethics burden)

1. Chick chorioallantoic membrane (CAM) assay for anti-angiogenic activity, or zebrafish toxicity and xenograft work. Both are cheaper and more ethically acceptable than rodent models.
2. Oxaliplatin-induced neuropathy or capecitabine hand-foot syndrome model in rodents. Behavioral tests (cold allodynia, von Frey) are cheap, and you can test a nutraceutical or repurposed drug.
3. DMH or AOM/DSS chemoprevention model with aberrant crypt foci counts and histology. It costs more and needs animal ethics approval, but it gives strong in vivo data.

E. Zero-wet-lab options (best if budget or time is very tight)

1. Pharmacovigilance data mining. Run disproportionality analysis on FAERS or VigiAccess for CRC drugs (bevacizumab, cetuximab, regorafenib, capecitabine). Look for unexpected adverse events. Add a hospital prescription audit of drug-drug interactions (for example capecitabine with warfarin or PPIs).
2. Bioinformatics pipeline. Use TCGA-COAD, GEO, GEPIA, cBioPortal and STRING to find hub genes, then dock candidate drugs. Even a small qPCR validation of one or two genes makes it much stronger.

How to make any of these look unique

• Combine an unusual input with a standard model. Examples: local fermented food, agro-waste, or an old cheap drug, tested in a spheroid model.
• Add a mechanism, not only an IC50. Include apoptosis, ROS, cell cycle, migration (scratch assay) and a few Western blot or qPCR markers.
• Use normal-cell selectivity. Compare against a normal colon line (CCD-841 or similar) and report a selectivity index.
• Include at least one computational step. Docking or ADMET prediction is free and looks rigorous.
• State the clinical link. For example: chemoresistance, side-effect reduction, or dose personalization.

Recent evidence context

• Immune checkpoint inhibitors in dMMR/MSI-high CRC (PMID 40591014 and PMID 40889731).
• Gut microbiome dysbiosis in CRC (PMID 40737178).
• Functionalized liposomes for colon-targeted delivery (PMID 38697444).

Next step

1. Your lab access (cell culture, animal house, or computer only).
2. Your time (months).
3. Your rough budget.

• Early-onset CRC and colibactin. A 2025 Nature genomic study of about 1,000 tumors linked colibactin-type mutation signatures (SBS88, ID18) from pks+ E. coli to cancers diagnosed before age 50. They often hit the APC gene (Labroots summary, The Pathologist). Researchers say causality still needs proof. That is a real gap, and nobody has a drug answer for it yet.
• GLP-1 receptor agonists and CRC. The 2026 PubMed hits are mostly letters debating the evidence (PMID 42608207, PMID 41698477). The question is unsettled.
• Ferroptosis in CRC. It now has a 2017 to 2025 bibliometric map (PMID 42890516), so a plain "compound X induces ferroptosis" project is crowded. Pair it with something unusual or skip it.

A. Early-onset CRC and environment (newest, strongest "wow")

1. Blocking colibactin. Test whether natural compounds or repurposed drugs reduce colibactin damage.
   • In silico: dock candidates against ClbP, the peptidase that activates colibactin.
   • In vitro: co-culture a pks+ E. coli with HT-29 cells and read out DNA damage (γH2AX staining or comet assay) or the "megalocytosis" cell enlargement assay.
   • Check your biosafety level before using any pks+ strain. The computational half alone is still a good project.
2. Food additives and the colon barrier. Expose Caco-2 or HT29-MTX cells to emulsifiers (CMC, polysorbate-80), titanium dioxide or artificial sweeteners. Measure barrier leakiness (TEER or a dye-permeability test) and see whether it changes 5-FU uptake or sensitivity. Ultra-processed food is a leading early-onset suspect.
3. Microplastics and mucus (high-risk, high-reward). Expose mucus-producing cells to polystyrene nanoparticles. The idea is that they weaken the mucus layer. The evidence is thin (I found only secondary web sources), so frame it as a hypothesis test.

B. Make cancer cells more vulnerable (cheap cell work, clear mechanism)

1. Tumor acidity and ion trapping. Tumors are acidic, which changes how weak-base drugs (irinotecan, doxorubicin) enter cells. Compare IC50 at pH 6.5 vs 7.4, then test whether bicarbonate or a proton pump inhibitor shifts it. It's very cheap and links pharmacokinetic theory (pH partition) to cancer biology.
2. Fasting-mimicking culture plus chemotherapy. Use low-glucose, low-serum medium and compare cancer cells with a normal colon line. The idea is that cancer cells lose stress resistance while normal cells keep it. This tests "differential stress resistance" cheaply.
3. Chemotherapy-induced senescence and senolytics. Induce senescence with oxaliplatin or 5-FU (SA-β-gal staining), then clear senescent cells with fisetin or quercetin. It addresses relapse and is cheap.
4. Immunogenic cell death in MSS tumors. Most CRCs are microsatellite stable ("cold") and do not respond to checkpoint inhibitors. Test whether a repurposed drug or oxaliplatin combination makes cells release HMGB1 and ATP and show surface calreticulin. This is an immunotherapy angle without needing immune cells.
5. Fibroblast-conditioned medium and resistance. Culture CRC cells in medium from a fibroblast line (CCD-18Co or NIH3T3) and test whether it protects them from 5-FU. Then reverse the protection with a drug. This is a simple tumor-microenvironment model.

C. AI and in silico (nearly free, looks advanced)

1. QSAR and virtual screening of Indian medicinal plants. Train a model on public cytotoxicity data (ChEMBL, NCI-60), screen the IMPPAT phytochemical database, and test the top 2 or 3 hits in vitro. This is "AI-guided natural drug discovery" with a validation step.
2. Food-derived anticancer peptides. Predict peptides from milk, soy, moringa or fish-waste proteins by simulated digestion (BIOPEP, PeptideCutter), then screen with AntiCP and ToxinPred. Validate with a real pepsin-pancreatin hydrolysate on cells. This is unusual and cheap.

D. Clinical safety and real-world (low lab cost, strong relevance)

1. 5-FU intestinal mucositis model. Induce diarrhea and gut damage in rats or mice with a single 5-FU dose, then test a repurposed or nutraceutical protector. Measure body weight, diarrhea score, MPO and histology.
2. Herb-drug interaction with capecitabine. Test turmeric, ashwagandha or giloy (all widely taken by patients), by docking and metabolism-enzyme assays, or a rat PK study if HPLC is available.
3. GLP-1 agonists and CRC in your hospital. A retrospective study in diabetic patients: do GLP-1 RA users differ in polyp or CRC rates vs other antidiabetic users? It needs ethics approval and records access. It's a hot, unsettled question, but it has confounding issues, so say so.
4. Pharmacoeconomics of DPYD testing. Build a cost-effectiveness (Markov) model of genotype-guided fluoropyrimidine dosing for your country. It needs only a computer, and it's easy to defend.
5. Pharmacist awareness and screening survey on early-onset CRC symptoms among young adults. It is easy and low cost, and externals value the public-health link. It has less pharmacology depth, so pair it with one of the others.

Best "two-layer" combinations for an external examiner

Cautions

• Check novelty yourself. I searched only trends, not every exact title. Search PubMed for your final combination and keep the closest 3 papers for your viva.
• Control for normal cells. Always include a normal colon line and report a selectivity index.
• Avoid overclaiming. Say "in vitro evidence suggests", not "cures".
• Ethics and biosafety. Idea 1 (bacteria), 11 and 12 (animals) and 13 (patient data) all need approvals.

CRC cell line panel (the base for most ideas)

Ideas

My top 5 for impact and ease

Cautions

• I did not run a full novelty search for each exact title. Do that on PubMed before you pick one.
• State cell-line genotypes with a source, and always include a normal colon line.
• Use the textbook molecular pathway in your introduction: normal epithelium, then APC, KRAS, p53 changes and the CIN/MSI/CIMP routes.

1. Resistance reversal. Drug X makes 5-FU-resistant colorectal cancer (CRC) cells respond again, through one named pathway.
2. Barrier protection. Drug X protects normal gut epithelium through one named marker.
3. Selectivity (implied). The protection must not shield the tumor.

What the literature supports

• Resistance mechanisms include thymidylate synthase (TYMS) overexpression, drug efflux by ABC transporters, EMT, autophagy and cancer stemness (review of 5-FU resistance, Biology 2021 review). Long-term 5-FU-resistant colon cells showed EMT, apoptosis evasion and autophagy (PMID 36067803). Rac1 signaling is another proposed resistance route (PMID 39513883).
• 5-FU is the main cause of gut mucositis in FOLFOX, and a 2026 review covers its pathogenesis and possible interventions. Repurposed or natural protectors in mice include losartan, ellagic acid and oxytocin. None was tested for resistance reversal in those papers.

Filled-in versions

Ready-to-paste statements

How to test each half

• Use HT-29 and HCT116, and ideally 5-FU-resistant sublines. Generating resistant lines by stepwise dose escalation takes about 3 months.
• A shortcut is to compare naturally less sensitive lines, but then you can't call it "reversal", only "sensitization".
• Mimic FOLFOX with 5-FU, folinic acid (leucovorin) and oxaliplatin at a fixed ratio. Report the combination index (CompuSyn), the IC50 shift and a reversal index (IC50 of resistant cells alone divided by IC50 with Drug X).
• Confirm the named pathway by qPCR or Western blot (for example TYMS, p21, cleaved caspase-3).

• Grow Caco-2 monolayers for about 21 days. Measure TEER, or use a FITC-dextran or Lucifer yellow permeability test if you lack an ohmmeter.
• Stain ZO-1 and occludin by immunofluorescence and quantify by Western blot.
• Expose the monolayer to 5-FU with and without Drug X.
• Do not use tight junction proteins in tumor cells as proof of protection. Tight junction proteins are expressed and often altered in cancer, so a barrier marker rising in tumor cells could mean the opposite of what you want.

• Show that Drug X does not reduce FOLFOX killing in cancer cells, and report a selectivity index against a normal colon line.
• This matters most for antioxidant-type protectors such as ellagic acid, which could blunt chemotherapy.

• Use a single-dose 5-FU mucositis model in mice or rats. Measure body weight, diarrhea score, villus height and crypt depth, MUC2 (PAS or Alcian blue), serum FITC-dextran leak and ZO-1.
• This needs ethics approval. A cheaper tumor-side option is the chick CAM assay.

Cautions

• Oxaliplatin's main toxicity is neuropathy, not gut injury, so frame the barrier claim around 5-FU.
• Leucovorin works by stabilizing the 5-FU/TYMS complex. Include it in your combination so the model is a true FOLFOX, not 5-FU plus oxaliplatin only.
• Claim "in vitro evidence suggests", not "reverses resistance in patients".
• I searched for the general concepts, not every drug-pathway pair. Run a final PubMed search on your chosen pair before committing.

Scoring the idea as written (sodium butyrate, UDCA or losartan version)

What decides whether you get 6 or 9

• Testing only cytotoxicity (IC50) and no mechanism.
• Using only one cell line and no normal colon control.
• Claiming "reverses resistance" without resistant cells.
• Using barrier markers in tumor cells as evidence of protection.
• Leaving leucovorin out of the FOLFOX mimic.
• Using drug concentrations that cannot be reached in patients.
• Overclaiming ("cures resistance" instead of "sensitizes in vitro").

• 5-FU-resistant sublines with a documented IC50 shift.
• Combination index and reversal index, with both calculated properly.
• A normal-cell selectivity index.
• A mechanism confirmed in at least two readouts (qPCR plus Western blot).
• A Caco-2 barrier assay (TEER or permeability) alongside the protein markers.
• A reasoned dose rationale.
• Honest limitations, such as no in vivo data and limited cell lines.

Likely viva questions to prepare for

• Why does protecting the gut not also protect the tumor?
• Why did you choose that pathway, and what would disprove it?
• How did you decide the drug concentration, and is it reachable in humans?
• Why is Caco-2 a fair barrier model, and where does it fall short?
• What does leucovorin do in FOLFOX?