TL;DR
You’ll find 60 biology research topics below, each with the equipment you need, how long it takes, and the exact tool or public dataset you’d use to run it. Twenty-seven need no lab at all. Pick one and start reading this week. You can explore any of these topics with Horizon, one-on-one with a professor or PhD mentor for a full term. You write a roughly 5,000-word paper, and it’s yours to submit to journals or competitions. It’s fully online; cohorts start multiple times a year.
Most lists of biology research topics are written for people who need an essay title. You’ll see “the ethics of genetic engineering” or “the human immune system,” and neither one tells you what you’d actually do on a Tuesday afternoon.
That’s the problem this page tries to fix. Every topic below is a question you can test, with a stated method, a stated time cost, and a note on what equipment it needs. Some need nothing but a laptop. Some need a bag of seeds and a windowsill. Only seven need a school lab, and none of them need a university lab you don’t have access to.
You don’t have to read the whole thing. Jump to the field you like, find two or three questions that make you curious, then go read what’s already been published on them. If you want the mechanics of running a project once you’ve chosen, we covered that separately in 8 tips on how to do biology research in high school. And if biology isn’t quite it, there’s a broader list of 49 research topics and a STEM research topics page too.
What makes a biology research topic doable?
There are four things that can make or break your research topic. Nail these, and the rest is just putting in the work.
- Scope: A solid topic keeps it simple: change one thing, measure one thing. For example, “How does climate change affect biodiversity?” That’s not a project; it’s a whole field! But “Does leaf litter break down faster in shaded soil or sunny soil?” Now that’s a project you can actually finish. If you can’t sum up your question in one sentence that includes a number, it’s still too big. Make it specific and measurable.
- Equipment: Be realistic about what you can actually use. If a spectrophotometer is locked away in a classroom, it’s not really available. On the other hand, your phone camera and ImageJ can measure leaf area, colony size, root length, and growth rate for free, and that combo is probably the most useful tool you’ll have.
- Data access: Some of the strongest projects here use data that’s already out there. Public databases offer gene expression runs, protein structures, species sightings, ocean temperatures, and clinical outcomes. Computational projects aren’t lesser; in fact, they’re often the best way for high schoolers to work with real data.
- Approval: Anything involving other people needs review before you start. For science fairs linked to ISEF, you’ll need to get approval from an Institutional Review Board before you recruit anyone or collect responses. Projects that use only public data, or just observe people in public with no interaction, are usually exempt. Figure this out at the very beginning.
How to choose and narrow your biology research topic?
- Pick the field before you pick the question. Choose the one you’d read about for fun. You’ll be living with this for months, and interest is the only thing that survives week six.
- Read three recent papers in that field. Not the whole paper. Read the abstract, then the last paragraph of the discussion, where researchers say what’s still unknown. That paragraph is a topic generator.
- Write your question with a variable in it. Take a broad interest and force it into the shape “does X change Y in Z.” Broad: plant stress. Shaped: does salt concentration change germination rate in Brassica rapa.
- Check you can measure Y. If you can’t name the instrument, the units and the number of samples, you don’t have a method yet. Count, time, weigh, or photograph and measure in software. Those four cover most of this page.
- Shrink it until it fits your calendar. Take your question and cut it in half. Then cut it again. A small finished study beats a big abandoned one every single time.
- Find someone who can poke holes in it. A teacher, a local grad student, a mentor. You want one person who’ll tell you your control group is wrong before you run the experiment, not after. If you don’t have that person nearby, here are seven ways to find a research lab, and 10 ways to get research opportunities in high school.
- Decide where it’s going before you start. A science fair, a research competition, a journal, or a conference each want different things. Knowing the destination shapes the project.
What are biologists actually researching right now?
Choosing a topic connected to current research makes your project both easier to write about and more compelling to judges. Below are some of the most active areas in biology as of 2026.
- Protein structure prediction has accelerated research progress. Structures that once took laboratories years to determine can now be predicted from amino acid sequence in minutes using free online tools. Several of the topics below incorporate this technology.
- Single-cell sequencing is another major advance. Researchers can now measure gene activity cell by cell, rather than averaging across entire tissues. Much of this data is publicly available for download and analysis.
- The microbiome—the collection of microorganisms living in and on organisms—remains a highly active research area, particularly its links to immunity and neurological function. Many questions in this field can be explored through culturing experiments or analysis of public datasets.
- CRISPR technology has expanded from gene knockout to precise base and prime editing, enabling single-nucleotide modifications. While gene editing requires specialized facilities, computational modeling of CRISPR targets is accessible for independent projects.
- Environmental DNA (eDNA) has transformed ecology by allowing researchers to identify organisms in an environment simply by sequencing DNA from water or soil samples. Some eDNA datasets are publicly available for independent analysis.
Which group should you start with?
| Choose this group if | Group | What you’ll need | How many topics |
|---|---|---|---|
| You like puzzles, code or data more than pipettes | Genetics and molecular biology | A laptop, mostly | 9 |
| You want visible results fast and can get agar plates | Microbiology and infectious disease | Home setup or school lab | 9 |
| You have a windowsill, some seeds and six weeks | Plant biology and agriculture | Kitchen counter | 8 |
| You’d rather be outside than indoors | Ecology and climate | Field notebook and a phone | 9i |
| You’re headed toward medicine or neuroscience | Human health and neuroscience | Home setup, some need approval | 9 |
| Water, coasts and climate are your thing | Marine and freshwater biology | Kitchen counter or public data | 8 |
| You already code, or want an excuse to learn | Computational biology | A laptop only | 8 |
If your interest sits outside biology, Horizon has physics research opportunities, disease research project ideas, and a bioinformatics programs list worth a look.
Genetics and molecular biology research topics
Nearly all of these run on a laptop. The databases are free, the tools are browser-based, and nobody will ask your age.
1. How closely related are two species, measured by a single gene?
What you need: No lab needed Time: 1 to 2 weeks Difficulty: Beginner
Pull the same gene sequence for several species from NCBI GenBank, align them with BLAST, and calculate percent identity for each pair. Then compare your ranking against the accepted family tree. The fun part is where they disagree. Different genes evolve at different speeds, and your ranking will show it.
2. Does a single point mutation destabilize a protein?
What you need: No lab needed Time: 2 weeks Difficulty: Intermediate
Download a structure from the Protein Data Bank, visualize it in PyMOL or the free Mol* viewer, then compare the wild type against a known disease mutation. See whether the changed residue sits buried in the core or on the surface. Buried changes tend to matter far more, and showing that pattern across several mutations is a full project.
3. Which genes change most between healthy and diseased tissue?
What you need: No lab needed Time: 3 weeks Difficulty: Intermediate
NCBI GEO holds thousands of free gene expression datasets. Pick one comparing healthy and diseased samples, calculate fold change for each gene, and build a volcano plot in Excel, R or Python. Your finding is the short list of genes that shift most, and what’s already known about them.
4. How fast does a gene mutate across related species?
What you need: No lab needed Time: 2 weeks Difficulty: Intermediate
Align the same gene across several related species using Clustal Omega, then calculate percent divergence. Compare a gene under strong selection against one that isn’t. You’re measuring the molecular clock, and the difference between the two genes is the whole story.
5. Does genetic drift hit small populations harder?
What you need: No lab needed Time: 1 week Difficulty: Beginner
Simulate allele frequencies over generations with random sampling in Excel or Python, running populations of 10, 100 and 1,000. Track how many runs lose an allele entirely. It’s the cleanest way to see why conservation biologists worry so much about small populations.
6. Can you map a gene regulatory network from public interaction data?
What you need: No lab needed Time: 3 weeks Difficulty: Advanced
STRING gives you protein interaction networks for free. Build a network around a gene you care about, then calculate node degree to find the hubs. Hub genes are often the ones that turn out to be drug targets, and comparing your hubs against known targets makes a strong result.
7. Which CRISPR guide RNA targets a gene most cleanly?
What you need: No lab needed Time: 2 weeks Difficulty: Intermediate
You can’t edit genes at home, but you can design the edit. Use CRISPOR to score candidate guide RNAs for a target gene, then compare on-target efficiency against predicted off-target hits. The trade-off between the two is a real problem working scientists argue about.
8. Do bacteria from different environments carry different resistance genes?
What you need: No lab needed Time: 3 weeks Difficulty: Intermediate
Search NCBI for antibiotic resistance gene sequences across bacterial genomes from soil, water and clinical samples. Count how often each resistance type appears in each environment. If clinical isolates carry more, that supports the case that our use of antibiotics drives the pattern.
9. Which fruit gives up the most DNA?
What you need: Home setup Time: 1 week Difficulty: Beginner
Extract DNA from strawberries, bananas, kiwis and onions using dish soap, salt and cold rubbing alcohol. Photograph the precipitate and measure the visible mass in ImageJ for a rough comparison. Strawberries are octoploid, meaning eight copies of each chromosome, and testing whether that shows up in your yield is the actual experiment.
Microbiology and infectious disease research topics
Bacteria grow fast, which means results in days instead of months. Work only with non-pathogenic strains, keep your plates sealed, and never open a plate you’ve grown from an environmental swab.
10. Does hand sanitizer actually beat soap and water?
What you need: Home setup Time: 2 weeks Difficulty: Beginner
Swab hands before and after each cleaning method, streak onto prepared agar plates, then count colonies after 48 hours. Run at least five trials per condition so you can test the difference statistically. The result often surprises people, since contact time matters more than the product.
11. How does antibiotic concentration change a bacterial growth curve?
What you need: School lab Time: 3 weeks Difficulty: Intermediate
Grow a non-pathogenic strain across a range of antibiotic concentrations and read optical density on a spectrophotometer every hour. Plot growth curves and find the concentration where growth stops. That value has a name, the minimum inhibitory concentration, and it’s what clinical labs measure every day.
12. Which surfaces in your house grow the most bacteria?
What you need: Home setup Time: 2 weeks Difficulty: Beginner
Swab doorknobs, phone screens, kitchen sponges and light switches, plate them, and count colonies. Then rank surfaces by colony count and by how often people touch them. The gap between the two rankings is your finding. Sponges usually win by a lot.
13. Do two bacterial strains suppress each other when grown together?
What you need: School lab Time: 3 weeks Difficulty: Intermediate
Culture two harmless strains separately, then together, measuring growth in each condition. If the co-culture grows less than the sum of the two alone, you’re seeing competition. It’s a small version of what happens in your gut every day.
14. What nutrient conditions make biofilms thickest?
What you need: School lab Time: 3 weeks Difficulty: Advanced
Grow biofilms in microplates under different nutrient levels, stain with crystal violet, and read absorbance. Compare conditions with ANOVA. Biofilms are why some hospital infections are so hard to clear. So this links straight to medicine.
15. Can an SIR model predict a real outbreak curve?
What you need: No lab needed Time: 2 weeks Difficulty: Intermediate
Build a Susceptible-Infected-Recovered model in Python or a spreadsheet, then fit it to published case data from a real outbreak. Tune the transmission rate until the curves match. What you learn is where the model breaks, which is usually the moment behavior changes.
16. How much UV exposure does it take to kill bacteria?
What you need: School lab Time: 2 weeks Difficulty: Beginner
Expose plated bacteria to a UV source for increasing lengths of time, then count surviving colonies. Plot survival against exposure and find the curve’s shape. It’s rarely a straight line, and explaining why is the interesting part.
17. Which natural preservatives actually stop mold?
What you need: Home setup Time: 3 weeks Difficulty: Beginner
Treat identical bread slices with salt, vinegar, honey, cinnamon and a control, seal them, and photograph daily. Measure mold coverage in ImageJ rather than guessing by eye. Quantifying an image is what separates this from a fifth grade project.
18. Does urban soil hold less microbial diversity than rural soil?
What you need: School lab Time: 4 weeks Difficulty: Intermediate
Collect soil from several urban and rural sites, plate serial dilutions, then classify colonies by morphology and calculate a Shannon diversity index for each site. Compare the groups statistically. Soil microbes move nutrients through the ground. So losing them shows up in the plants above.
Plant biology and agriculture research topics
Plants are the friendliest organisms for a first project. They’re cheap, they’re legal, nobody needs to approve your protocol, and a windowsill is a valid lab.
19. At what salt concentration do seeds stop germinating?
What you need: Home setup Time: 3 weeks Difficulty: Beginner
Germinate Brassica rapa or radish seeds across a gradient of salt solutions, recording germination rate and root length. Measure roots from photographs in ImageJ for accuracy. Soil salinity is a live agricultural problem, and finding the threshold where germination collapses is a real result.
20. Which color of light drives photosynthesis fastest?
What you need: School lab Time: 2 weeks Difficulty: Beginner
Count oxygen bubbles from Elodea under red, blue, green and white filters at a fixed distance. Then compare your results to the chlorophyll absorption spectrum. Green light should perform worst, and confirming that experimentally is more satisfying than reading it.
21. Does caffeine slow down seed germination?
What you need: Home setup Time: 3 weeks Difficulty: Beginner
Soak seeds in a range of caffeine concentrations, then track germination timing and root length. Build a dose-response curve. Some plants make caffeine to suppress competitors, so you’re testing an actual ecological hypothesis, not just a kitchen experiment.
22. How much does auxin concentration change stem growth?
What you need: Home setup Time: 4 weeks Difficulty: Intermediate
Apply rooting hormone at several dilutions to seedlings and measure stem elongation over three weeks. Plot growth against concentration. The curve usually peaks and then falls, since too much auxin inhibits growth, and catching that turnaround is the win.
23. Do companion plantings actually help each other grow?
What you need: Home setup Time: 6 weeks Difficulty: Beginner
Grow basil and tomato together, then separately, with everything else held constant, and compare biomass at the end. Gardening advice about companion planting is mostly folklore with thin evidence behind it. Testing one pairing properly is a genuine contribution.
24. How much carbon can a fast-growing plant store in one season?
What you need: Home setup Time: 8 weeks Difficulty: Intermediate
Grow several species, harvest, dry to constant mass, and apply standard carbon conversion factors to estimate sequestration per plant. Compare species on a per-gram and per-day basis. It scales up to a real question about which crops are worth planting for carbon.
25. Does fungal inoculation build bigger roots?
What you need: Home setup Time: 8 weeks Difficulty: Intermediate
Grow seedlings with and without commercial mycorrhizal inoculant, then wash, photograph and measure root mass and branching. Most plants trade sugar to fungi for better nutrient uptake. Whether a store-bought inoculant delivers on that is a fair question.
26. Can you detect drought stress before you can see it?
What you need: Home setup Time: 4 weeks Difficulty: Advanced
Photograph plants daily under different watering schedules and analyze leaf color channels in ImageJ to catch shifts your eye misses. Compare the day your measurement detects stress against the day it becomes visible. Early detection is exactly what agricultural drone imaging is built to do.
Ecology, environment and climate research topics
Field ecology is the cheapest research there is. Your equipment is a notebook, a phone and some patience, and the datasets that back it up are all free.
27. How much does temperature vary across your city?
What you need: Home setup Time: 4 weeks Difficulty: Beginner
Log temperature at the same times each day across a park, a parking lot, a tree-lined street and an open plaza. Compare daily ranges, not just averages. Urban heat islands are well documented at city scale, but mapping them street by street is local data nobody else has.
28. Are pollinators showing up earlier than they used to?
What you need: No lab needed Time: 3 weeks Difficulty: Intermediate
Pull decades of observation records for a pollinator species from GBIF or iNaturalist, extract first-sighting dates by year, and run a regression against time. If first sightings are creeping earlier, you’ve measured phenological shift. This is real climate science using a free database and a spreadsheet.
29. Do microplastics change survival in small aquatic organisms?
What you need: School lab Time: 4 weeks Difficulty: Intermediate
Expose Daphnia or brine shrimp to a range of microplastic concentrations and track survival and reproduction. Analyze with regression or ANOVA. Most published work uses concentrations far above what’s in real water, so testing realistic levels is a smarter design.
30. How different are two habitats a hundred meters apart?
What you need: Home setup Time: 3 weeks Difficulty: Beginner
Run quadrat surveys in two nearby microhabitats, count species, then calculate Shannon and Simpson diversity indices for each. Compare them statistically. Edge effects mean two spots that look similar often aren’t, and finding that is the point.
31. Does light pollution change insect activity at night?
What you need: Home setup Time: 4 weeks Difficulty: Intermediate
Set light traps at sites with different night-time brightness and count insects by group across several nights. Use a phone light meter for brightness readings. Insect decline is one of the bigger open questions in ecology. Light is a suspect, and there’s very little local data on it.
32. What makes leaf litter break down faster?
What you need: Home setup Time: 8 weeks Difficulty: Beginner
Bury mesh bags of dried, pre-weighed leaves in different soil conditions, then recover and reweigh them every two weeks. Plot mass loss over time. Decomposition rate controls how fast carbon returns to the atmosphere, so a small backyard study connects to a very big cycle.
33. What’s the carrying capacity of a local population?
What you need: No lab needed Time: 2 weeks Difficulty: Intermediate
Fit logistic growth models to published population data for a local species, estimate carrying capacity, then test how the estimate shifts under different resource assumptions. Modeling projects like this need no permits and no field season, which matters if you’re starting in November.
34. Are bird ranges moving north?
What you need: No lab needed Time: 3 weeks Difficulty: Intermediate
Download occurrence records for a few bird species from GBIF or eBird, calculate mean latitude of sightings per year, and regress against time. Compare a migratory species against a resident one. If only one is moving, that difference needs explaining, and that’s your discussion section.
35. How much does a toxin concentrate as it moves up a food chain?
What you need: No lab needed Time: 2 weeks Difficulty: Intermediate
Use published contaminant data across trophic levels to calculate biomagnification factors, then model how concentration builds from producer to top predator. Mercury in fish is the classic case and the data is public. Working out how many steps it takes to reach an unsafe level makes the abstract concrete.
Human health, physiology and neuroscience research topics
Read this before you start any of these. If your project collects data from other people, including surveys, it needs Institutional Review Board approval before you recruit anyone. Using only yourself as a subject, or working with public datasets, avoids that requirement.
36. How fast does your heart rate recover after different workouts?
What you need: Home setup Time: 3 weeks Difficulty: Beginner
Record heart rate recovery curves after light, moderate and hard exercise using a fitness tracker or manual pulse counts, then calculate the slope of the first minute. Recovery slope is a recognized fitness marker used by cardiologists. Running it on yourself across several weeks keeps it out of review board territory.
37. Does one bad night of sleep measurably slow you down?
What you need: Home setup Time: 4 weeks Difficulty: Intermediate
Run a simple reaction time test at the same hour each day and log sleep duration alongside it, then look for a correlation. Needs IRB approval if you test anyone besides yourself. Reaction time is noisy, so you’ll need far more trials than feels necessary, and learning that is part of the project.
38. How quickly do you get good at something new?
What you need: Home setup Time: 3 weeks Difficulty: Beginner
Pick a motor task like mirror drawing or a typing pattern, run repeated trials, and plot the learning curve. Fit it and calculate the rate constant. Learning curves almost always flatten rather than stopping, and showing that in your own data is neat.
39. Does screen time before bed change how well you sleep?
What you need: Home setup Time: 4 weeks Difficulty: Intermediate
Track sleep quality scores against evening screen use with a wearable or a sleep diary. Requires IRB approval for any participant other than yourself. The confounders here are brutal. People who scroll late differ in other ways too, and naming those honestly makes your paper stronger.
40. How does breathing rate respond to activity?
What you need: Home setup Time: 2 weeks Difficulty: Beginner
Count breaths per minute before, during and after controlled activity, then plot the recovery curve and calculate return-to-baseline time. Compare across fitness levels if you have approval to test others. It’s simple, it’s cheap, and the data is remarkably clean.
41. Can you model how immunity builds with repeated exposure?
What you need: No lab needed Time: 3 weeks Difficulty: Intermediate
Build a spreadsheet or Python model of antibody levels over time, varying exposure timing and dose. Compare your output against published vaccination schedules. If your model explains why boosters are spaced the way they are, you’ve landed something worth writing up.
42. What does open brain imaging data show about language?
What you need: No lab needed Time: 4 weeks Difficulty: Advanced
OpenNeuro hosts free brain imaging datasets. Download a language task study and compare activation patterns across participants. It’s the hardest project on this page and needs real patience with the tools, but almost no high schooler ever touches neuroimaging data.
43. Does caffeine actually improve attention?
What you need: Home setup Time: 3 weeks Difficulty: Intermediate
Run attention tests before and after caffeine using a blinded design where possible, with yourself as the subject unless you have IRB approval. Test at consistent times of day. Tolerance means a daily coffee drinker and a first-timer will give you completely different curves.
44. What does public cancer genomics data say about survival?
What you need: No lab needed Time: 4 weeks Difficulty: Advanced
Use the Genomic Data Commons to compare gene expression against patient survival for one cancer type. Build survival curves and test whether a gene’s expression level separates the groups. This is exactly what cancer biologists do, using exactly the same data.
Marine and freshwater biology research topics
You don’t need an ocean. A jar on a windowsill models more than you’d expect, and the satellite and survey data covering real oceans is entirely free.
45. How does temperature change algal growth?
What you need: Home setup Time: 4 weeks Difficulty: Beginner
Grow algae cultures at several temperatures and track density by turbidity or by photographing samples against a standard. Plot growth rate against temperature. Warming water is driving bloom changes worldwide, and your curve is a small version of that.
46. At what salinity do brine shrimp stop hatching?
What you need: Home setup Time: 2 weeks Difficulty: Beginner
Hatch brine shrimp eggs across a salinity gradient, counting hatch rate and timing. Find the optimum and the limits. Brine shrimp put up with salt levels that would kill almost anything else. So mapping where they finally give out makes a clean dose-response study.
47. Can you trigger an algal bloom in a jar?
What you need: Home setup Time: 4 weeks Difficulty: Intermediate
Add fertilizer at increasing concentrations to identical pond water samples and track turbidity and dissolved oxygen over time. Watch what happens to oxygen after the bloom peaks. The crash is the part that kills fish, and seeing it happen in a jar explains dead zones better than any diagram.
48. Where are ocean temperatures rising fastest?
What you need: No lab needed Time: 3 weeks Difficulty: Intermediate
Download sea surface temperature records from NOAA for several regions and calculate warming rates per decade. Compare polar, temperate and tropical sites. The rates differ a lot by region. And explaining why is real oceanography.
49. Does acidified water weaken shells?
What you need: Home setup Time: 3 weeks Difficulty: Beginner
Soak cleaned shells or chalk in solutions at different pH levels, measuring mass loss over time. Track pH daily, since it drifts. Vinegar is far stronger than real ocean acidification, so be careful how you scale your conclusion, and saying so out loud is good science.
50. How does cloudy water affect aquatic plants?
What you need: Home setup Time: 4 weeks Difficulty: Beginner
Grow aquatic plants under different turbidity levels created with suspended sediment, measuring growth and oxygen output. Mud washing off building sites and farms is a big stress on rivers. Putting a number on the growth cost at each level is the useful part.
51. What does environmental DNA reveal about a river?
What you need: No lab needed Time: 3 weeks Difficulty: Advanced
Published eDNA studies release their sequence data. Download a dataset, run the sequences through BLAST to identify species, and compare the detected community against traditional survey records for the same river. Where the two methods disagree is the finding.
52. Are fish populations shifting where they live?
What you need: No lab needed Time: 3 weeks Difficulty: Intermediate
Use NOAA trawl survey data to track mean latitude or depth of a commercially important species over decades. Regress position against year and against water temperature. Fisheries management depends on answers to exactly this question, and the data is open to anyone.
Computational biology and bioinformatics research topics
Every topic here needs a laptop and nothing else. If you’re starting in the middle of winter with no lab access, this is your group.
53. Can a model spot a diseased leaf from a photo?
What you need: No lab needed Time: 4 weeks Difficulty: Advanced
Train an image classifier on a public plant disease dataset using Python with scikit-learn or TensorFlow, then report accuracy and confusion matrix. Test what happens when you feed it photos you took yourself. Models that score 95% on clean data often fall apart on real phone photos. Writing up that gap makes a better paper than the accuracy score.
54. Does bacterial growth really follow a logistic curve?
What you need: No lab needed Time: 2 weeks Difficulty: Beginner
Fit both exponential and logistic models to published bacterial growth data and compare fit using R squared and residual plots. Check where each model fails. Textbooks present logistic growth as settled, and finding where it doesn’t hold is a legitimate small result.
55. Can you predict what a protein does from its sequence alone?
What you need: No lab needed Time: 3 weeks Difficulty: Intermediate
Take proteins of known function, run them through domain prediction tools like InterPro, and score how often the prediction matches the known annotation. Then test it on proteins nobody has studied much. You’re measuring how far we can trust automated labels, and that’s a question the field argues about.
56. Does a genetic toggle switch survive noise?
What you need: No lab needed Time: 3 weeks Difficulty: Advanced
Model a two-gene toggle switch in Python, then add random noise and measure how often the system flips states unprompted. Map stability against noise level. Synthetic biology circuits fail in living cells for exactly this reason, and modeling it explains why.
57. Can you rebuild the tree of life from scratch?
What you need: No lab needed Time: 3 weeks Difficulty: Intermediate
Pull sequences for 15 to 20 species, align them, build a phylogenetic tree with MEGA, then compare your tree against the accepted one. Try it with a different gene and see whether the tree changes. When two genes give two trees, something interesting happened in evolutionary history.
58. How well do temperature coefficients predict metabolism?
What you need: No lab needed Time: 2 weeks Difficulty: Intermediate
Gather published metabolic rate data across temperatures for several ectotherms, calculate Q10 coefficients, and test whether one value holds across the full temperature range. It usually doesn’t at the extremes. Finding where the rule breaks beats confirming that it works.
59. Could an existing drug treat a different disease?
What you need: No lab needed Time: 4 weeks Difficulty: Advanced
Cross-reference drug target databases against disease gene lists to find drugs whose targets overlap with a disease you didn’t expect. Rank candidates by overlap strength. Several approved drugs were found this way. And the databases are open to anyone.
60. Does a fancier model beat a simple one on clinical data?
What you need: No lab needed Time: 4 weeks Difficulty: Advanced
Take a public clinical dataset, build both logistic regression and random forest classifiers, and compare accuracy with proper cross-validation. Report interpretability alongside accuracy. Simple models win more often than the hype suggests. Showing that with your own numbers is a confident result.
Which of these topics work for science fairs and ISEF?
Science fairs are judged differently than school grades. Judges want a clear variable, real replication, honest statistics, and a student who understands the limitations of their own study.
ISEF runs 22 categories, and the biology-relevant ones are Cellular and Molecular Biology, Microbiology, Plant Sciences, Animal Sciences, Biochemistry, Biomedical and Health Sciences, Computational Biology and Bioinformatics, Earth and Environmental Sciences, and Translational Medical Science. You choose your own category, so pick the one whose judges will best understand what you did.
Here’s how the groups on this page map onto fair categories:
| If you pick from | Likely ISEF category | Watch out for |
|---|---|---|
| Genetics and molecular biology | Cellular and Molecular Biology (CELL), Computational Biology (CBIO) | Computational projects need heavier analysis to compete |
| Microbiology | Microbiology (MCRO) | Plate work needs a supervised lab and sealed plates |
| Plant biology | Plant Sciences (PLNT) | Replication counts, run more plants than feels needed |
| Ecology and climate | Earth and Environmental Sciences (EAEV) | Field data varies a lot, so sample across several days |
| Human health | Biomedical and Health Sciences (BMED) | Needs IRB approval before you start |
| Marine and freshwater | Earth and Environmental Sciences (EAEV) | Jar models need careful framing about what they represent |
| Computational biology | Computational Biology and Bioinformatics (CBIO) | Public data projects need an original question, not a tutorial |
The paperwork trips more students than the science does. Human participant projects need an Adult Sponsor Checklist, a Student Checklist, a research plan, an approval form, a Human Participants form, and signed informed consent, all before you collect anything. Studies using only public data or observations in public settings with no interaction are generally exempt.
If a fair is your destination, Horizon has lists of 15 science competitions, 15 research competitions, and research awards worth entering.
Frequently asked questions
No. Twenty-seven of the 60 topics on this page need nothing but a laptop, and 26 more run on a kitchen counter. Only seven need a school lab. Lack of lab access is the most common reason students give up on research, and it’s usually the wrong reason. If you do want lab access, here are seven ways to find a research lab.
That’s a separate skill and we wrote it up properly in 8 tips on how to do biology research in high school, which covers experimental design, controls, statistics, safety and the writing structure. Start there once your question is settled.
The topics here range from one week to eight. Most students underestimate the write-up, which often takes as long as the experiment. We broke down realistic timelines in how long it takes to write a research paper.
Yes, and several journals exist specifically for high school work. Start with 15 research journals for high school students, then read how to publish research in high school for the submission process. Nothing is guaranteed here. So treat publishing as a goal, not a plan.
Then pick from the computational groups, where your zip code decides nothing. Location is a real barrier for lab-based programs and no barrier at all for data-based research. Horizon works the same way, since it’s fully online: you’re matched with a professor or PhD mentor, you work one-on-one for a term, and you finish with a full research paper. Tuition is $6,950 per course, and need-based aid is available for households under $75,000. For other options, see free online research programs and online research mentorship programs.
The ones marked Beginner, and particularly topics 5, 9, 12, 19, 30, 38 and 46. All of them need minimal equipment and give you a measurable result inside three weeks. Starting small and finishing is worth far more than starting big and stalling.
They can, when the work is real and you can talk about it. We looked at the evidence in do summer research programs actually help with college admissions and what you can actually do with a high school research paper.
Start with one question this week
Pick one topic. Read three papers on it. Write your question in one sentence with a variable in it. That’s your week, and it’s further than most people get.
What most students are missing isn’t a topic, it’s someone who’ll tell them their design is wrong before they run it. Horizon Academic pairs you with a professor or PhD mentor for a full term, one-on-one, on a single research question you choose. You finish with a roughly 5,000-word research paper that’s yours to submit to journals or competitions, across 600+ research specializations including genetics, neuroscience, environmental health and bio-industry.
Cohorts start three times a year, it’s fully online, and applications are open now.
More from Horizon
Doing the research 8 tips on how to do biology research in high school · Key components of a high school research paper · How long does it take to write a research paper · How to format citations
Getting it published or entered How to publish research in high school · 15 research journals for high school students · 7 steps to turn your paper into a competition submission · Research conferences for high school students · 15 science competitions · 15 research competitions · Research awards
Finding a program or a lab 7 ways to find a research lab · 10 best ways to get research opportunities in high school · 15 best biology programs · 15 genetics summer programs · 15 bioinformatics research programs · 15 biotech internships · 15 life science summer programs · 15 molecular biology courses
If you need funding 11 free research mentorship programs · 14 grants for high school student research · 16 fully-funded STEM programs · Free online research programs
More topic ideas 49 research topics for high school students · STEM research topics · 15 disease research project ideas · Physics research opportunities
Image source: Horizon Academic Research Program