First day doing Bioinformatics

I'm an independent bioinformatics researcher (no institutional affiliation, no grant funding) working solo on public datasets and on-demand cloud compute (RunPod/Vast.ai for GPU, Hetzner/AWS spot for CPU). I have real prior bioinformatics/ML background — I'm not starting from zero — but I'm working with a hard budget ceiling of roughly €50-100 total per project and a timeline of days, not months.

I want to find genuinely OPEN problems: questions nobody has publicly answered yet, not replications of existing published work. I'm not looking for a "cool technique to try" — I already have a long list of those. I specifically want gaps: a real question, answerable with data I can actually get, that as far as you can tell hasn't been answered in the literature.

Hard constraints on any candidate:
- Data: fully public, no institutional login/access application required (rules out raw UK Biobank, dbGaP-controlled access, etc. — summary statistics and open repositories like GEO, SRA, ENA, IEU OpenGWAS, GWAS Catalog, PGS Catalog, ClinVar, IEDB, CELLxGENE Census, HPRC, NCBI Pathogen Detection are all fine)
- Compute: must run CPU-only or on modest/short GPU rental — nothing needing a compute cluster or a multi-day training run
- Budget: total cost including any cloud rental should realistically stay under €100, ideally under €30
- Timeline: a solo researcher should be able to get a first real result within roughly a week of focused work

Search across these source types for candidate gaps:
1. "Future work" / "limitations" sections of recent (last 1-2 years) papers in relevant journals (Bioinformatics, Nature Genetics, Genome Biology, Cell Genomics, PLOS Computational Biology) — these often name a specific unanswered question directly
2. Review/perspective articles that explicitly list open questions in a subfield
3. bioRxiv/medRxiv preprints proposing a method but only validating it on 1-2 datasets, leaving obvious "does this generalize to X" gaps
4. GitHub issues/discussions on widely-used bioinformatics tools (e.g. requests like "has anyone tried this on dataset Y" that went unanswered)
5. Public GWAS/expression/methylation/single-cell datasets that exist but appear to have no or minimal downstream analysis published against them (check citation counts / "cited by" on the dataset's original paper)
6. DREAM Challenges, Kaggle bioinformatics competitions, or CAMDA challenges with sub-problems nobody fully solved, or follow-up questions the organizers explicitly flagged as unresolved
7. Recent conference talks/preprints (e.g. from ISMB, RECOMB) that end with an open question rather than a finished result

For each candidate problem you find, report:
- The specific question (framed precisely enough that "yes/no" or a concrete number would answer it)
- Why it's plausibly still open — what you searched to confirm this, and what you found (or didn't find)
- What free/public data would be needed, and confirmation it's actually accessible (name the specific dataset/repository)
- Rough method needed to answer it (doesn't need to be novel — an established method applied to this specific unanswered question is exactly what I want)
- Estimated compute/budget/timeline given the constraints above
- Why the answer would matter to someone (clinical relevance, methodological insight, or a genuine "this changes how people should think about X")
- Your honest confidence this is really unaddressed (High/Medium/Low) — flag if paywalled work might already exist that you can't fully verify

Give me 8-10 candidates spanning a few different subfields (don't cluster them all in one area), ranked by (confidence it's genuinely open) x (feasibility within my constraints) x (how much the answer would actually matter). Be skeptical of your own candidates — I'd rather have 5 solid ones than 10 where half turn out to already be published somewhere you didn't check.