Yila
All use cases
Literature

From the lab’s direction to your own topic

Upload the lab’s representative papers and Yila extracts the long-term research object, perturbations, and core mechanisms, draws the mainline map with exactly where the evidence stops, then searches five years of outside progress, proposes candidate gaps in a three-layer evidence matrix, verifies each by reverse search, and filters down to three advisor-ready candidate topics that fit your resources.

These are representative papers from my lab’s research direction. First extract the lab’s long-term research object, perturbations, core mechanisms, and usual study designs. Do not propose new topics yet.

Extraction complete: the lab works on primary human airway smooth muscle cells, around glucocorticoid sensitivity, β2 agonists, and inflammatory mechanisms. Solid lines are limited to literature-backed relations; dashed lines mark inference still to be verified.

Draw a simple mainline map from these papers: research object, stimuli, response genes, inflammatory or functional outcomes, and where the evidence stops. Do not draw speculation as proven relations.

The mainline map and evidence matrix are ready: lab papers and the closest outside studies are aligned by object, stimulus, molecular mechanism, functional outcome, evidence tier, and open problems — reviews and primary studies labeled separately.

The mainline map and evidence matrix are ready: lab papers and the closest outside studies are aligned by object, stimulus, molecular mechanism, functional outcome, evidence tier, and open problems — reviews and primary studies labeled separately.
Propose candidate gaps from five angles: mechanism, research object, conditions, evidence type, and real-world application. Each gap must cite its supporting papers — never just “understudied”.

Each candidate gap comes with its evidence. For example R3: whether different inflammatory endotypes and patient subgroups can get mechanism-guided treatment — grounded in P06, E02, P05 and others, with the specific open questions listed instead of a bare “understudied”.

Each candidate gap comes with its evidence. For example R3: whether different inflammatory endotypes and patient subgroups can get mechanism-guided treatment — grounded in P06, E02, P05 and others, with the specific open questions listed instead of a bare “understudied”.
Reverse-search each candidate gap to check for existing direct studies. Remove or narrow the ones already done; mark insufficient evidence as pending verification.

Reverse verification complete: candidates with direct prior work were removed or narrowed, the under-evidenced are marked pending, and every surviving gap keeps its search record for traceability.

Here are our resources and constraints. Filter to 3 candidate topics by significance, continuity with the lab’s mainline, research increment, resource fit, timeline, and risk. Do not recommend anything clearly beyond our means.

The three candidates are compressed into a one-page advisor table, with a flowchart from lab accumulation to outside progress, research gaps, resource screening, and candidate topics — all framed as discussion proposals, not settled conclusions.

The three candidates are compressed into a one-page advisor table, with a flowchart from lab accumulation to outside progress, research gaps, resource screening, and candidate topics — all framed as discussion proposals, not settled conclusions.
evidence-matrix.xlsxEvidence matrix · Excel
research-mainline.pngMainline map · PNG
candidate-topics.csvCandidate topics · CSV
claim-evidence-map.mdClaim–evidence map · Markdown
pubmed-search-results.csvSearch records · CSV
Task completed

Yila task replay complete.

Start with this method