What is cell type annotation?

Cell type annotation is the process of identifying and labeling different cell types in single-cell RNA sequencing (scRNA-seq) data by analyzing gene expression patterns. It's a crucial step in single-cell analysis that assigns biological meaning to cell clusters.

How does AI improve scRNA-seq annotation?

AI-powered annotation leverages large language models trained on extensive biological literature to provide accurate, context-aware cell type identification. Multiple AI models work in consensus to reduce bias and improve annotation reliability compared to single-method approaches.

What file formats does mLLMCelltype support?

mLLMCelltype supports CSV, TSV, and Excel (.xlsx) formats containing marker genes identified from your scRNA-seq analysis. Upload your top differentially expressed genes per cluster, with files up to 16MB in size.

How accurate is automated cell type annotation?

mLLMCelltype achieves 94.2% average accuracy across diverse datasets, with 96.8% accuracy for major cell types. The multi-model consensus approach provides confidence scores for each annotation, helping identify results that may need manual review.

Is mLLMCelltype free to use?

Yes, mLLMCelltype is a free, open-source tool for the research community. Users need to provide their own API keys for the AI models they wish to use (GPT-4, Claude, Gemini, etc.).

🧬

{{ t('fileName') }}: {{ uploadedFile.name }}

{{ t('fileSize') }}: {{ formatFileSize(uploadedFile.size) }}

CSV TSV Excel

{{ column }}

{{ t('selectedModelsCount', { count: provider.selectedModels.length }) }}

{{ t('consensusTipTitle') }}

{{ t('selectedModels') }}: {{ selectedModelNames.join(', ') }}

{{ t('species') }}: {{ species }}

{{ t('tissue') }}: {{ tissue }}

{{ Math.round(progressPercent) }}% {{ getProgressPhaseLabel(progressDetails.phase) }}

{{ t('cluster') }}	{{ t('cellType') }}	{{ t('consensus') }}	{{ t('entropy') }}	{{ t('confidence') }}
{{ clusterId }}	{{ cellType }}	{{ (results.consensus_proportion[clusterId] * 100).toFixed(1) }}%		{{ getConfidenceLabel(results.consensus_proportion[clusterId]) }}

{{ t('welcomeTitle') }}

{{ t('featureMultiModel') }}

{{ t('featureConsensus') }}

{{ t('featureSecurity') }}

{{ t('uploadTitle') }}

{{ t('dragDropTitle') }}

{{ t('uploadingTitle') }}

{{ t('uploadSuccessTitle') }}

{{ t('supportedFormats') }}

{{ t('needSampleData') }}

{{ t('dataPreview') }}

{{ t('uploadError') }}

{{ t('configTitle') }}

{{ t('basicSettings') }}

{{ t('importApiKeys') }}

{{ t('modelSelection') }}

{{ provider.name }}

{{ t('advancedOptions') }}

{{ t('processingTitle') }}

{{ taskStatus }}

{{ t('processingError') }}

{{ t('resultsTitle') }}

{{ Object.keys(results.consensus).length }}

{{ selectedModelNames.length }}

{{ formatDuration(processingTime) }}

{{ t('annotationResults') }}

{{ toast.title }}