Preparing GraphSol inputs

You provide proteins as sequences — pasted in FASTA, fetched from UniProt IDs, or imported from a previous job. The one thing that most affects your result is sequence integrity: use standard amino acids and unique identifiers, and keep each sequence within the accepted length range so the graph features generate cleanly.

Webapp input methods

The .bloom* webapp provides three input methods for GraphSol protein solubility prediction:

Input method selection

FASTA sequences input (default)

• Format: Standard FASTA with header lines starting with >
• Multiline textarea: 6 rows for sequence entry
• Example placeholder:

  >protein1
  MRLVQLSRHSIAFPSPEGALREPNGLLALGGDLSPARLLMAYQRGIFPWFSPGDPILWWSPDPRAVLWPESLHISRSMK
  >protein2
  MQEIYRFIDDAIEADRQRYTDIADQIWDHPETRFEEFWSAEHLASALESAGFTVTRNVGNIPNAFIASFGQGKPVIALL

• Validation: Must include > header lines

UniProt IDs input

• Dynamic list: Add multiple UniProt IDs with "Add UniProt ID" button
• Format: Standard UniProt accessions (e.g., P00533)
• Auto-fetch: System retrieves sequences from UniProt database
• Validation: Alphanumeric characters and underscores only

From previous job

• Job Output Selector: Browse completed jobs with protein sequence outputs
• Data type: Requires protein_sequence output
• Compatible tools: BLAST results, ESMFold inputs, other protein analysis

Webapp parameters

Job options

Personalized settings

• Use Personalized Settings: Toggle for project-specific configuration
• Loads settings from Supabase when enabled

Submit button

• Label: "Predict Protein Solubility"
• Disabled when: Input validation fails (no sequences/IDs or invalid format)

Output information

The tool provides:

• Main results CSV with solubility predictions and confidence scores
• Quality metrics summary for feature generation assessment
• Detailed feature files organized by protein (PSI-BLAST, HH-Suite, contact predictions)

Protein sequence input methods in .bloom* platform

GraphSol accepts various protein sequence formats through .bloom*'s intuitive web interface. The tool provides real-time validation, sequence preview, and automatic format detection to ensure reliable solubility predictions.

Direct sequence entry methods

Single sequence input

FASTA format:

• Standard FASTA format with header line starting with ">"
• Descriptive headers for result tracking (e.g., ">MyProtein_001")
• Single protein sequence per submission
• Automatic validation and preview

Raw sequence:

• Plain amino acid sequence without headers
• Direct paste of sequence strings
• Automatic format detection and conversion
• Real-time sequence property calculation

Batch sequence processing

Multi-sequence FASTA files:

• Upload FASTA files containing multiple protein sequences
• Automatic parsing of headers and sequences
• Progress tracking for large sequence libraries
• Error reporting for individual sequence issues

Tab-separated format:

• Text files with <identifier>\t<sequence> format
• One sequence per line
• Automatic delimiter detection
• Support for custom identifiers

Sequence validation and preprocessing

Character validation

Amino acid alphabet:

• Only standard amino acids accepted: ACDEFGHIKLMNPQRSTVWY
• Automatic removal of whitespace, numbers, and special characters
• Case-insensitive processing (converted to uppercase)
• Warning for non-standard characters

Sequence cleaning:

• Automatic whitespace stripping
• Number and punctuation removal
• Invalid character filtering
• Sequence normalization for consistency

Length and size requirements

Sequence length limits:

• Minimum: 20 amino acid residues
• Maximum: 4,096 amino acid residues
• Optimal range: 50-2,000 residues for best performance
• Very short sequences: May have reduced prediction accuracy

File size constraints:

• Maximum file size: 25 MB per submission
• Compression support: Automatic gzip (.gz) detection
• Large datasets: Split into multiple jobs if needed
• Batch limits: Thousands of sequences supported

Sequence quality control

Identifier management

Unique identifiers:

• Each sequence must have a unique identifier
• Automatic duplicate detection and removal
• Warning messages for duplicate sequences
• Preservation of original identifiers in results

Identifier formatting:

• Descriptive but concise headers recommended
• Avoid special characters in identifiers
• Consistent naming conventions across batches
• Support for alphanumeric characters and underscores

Sequence integrity checks

Format validation:

• Proper FASTA header formatting
• No empty lines between headers and sequences
• Consistent sequence representation
• UTF-8 encoding requirement

Content validation:

• Amino acid composition analysis
• Sequence length verification
• Basic physicochemical property calculation
• Unusual composition warnings

Advanced input options

Integration with other tools

BLAST results import:

• Direct import from BLAST sequence search results
• Automatic extraction of hit sequences
• Preservation of alignment information
• Batch processing of homologous sequences

Sequence analysis workflows:

• Import from multiple sequence alignment outputs
• Connection to motif analysis results
• Integration with domain prediction tools
• Support for clustered sequence datasets

File handling features

Format conversion:

• Automatic conversion between input formats
• Standardization to internal representation
• Preservation of sequence metadata
• Error recovery for malformed inputs

Batch processing:

• Parallel validation of multiple sequences
• Individual error handling per sequence
• Progress monitoring for large files
• Partial success reporting for mixed-quality inputs

Input preparation workflows

Research laboratory workflow

1. Export sequences from your analysis software
2. Format as FASTA with descriptive headers
3. Validate locally using sequence analysis tools
4. Upload to .bloom through the GraphSol interface
5. Review validation results before submission

High-throughput screening

1. Prepare sequence libraries in batch FASTA format
2. Apply consistent naming conventions
3. Compress large files using gzip
4. Upload and validate through web interface
5. Monitor batch processing progress
6. Review results and plan next steps

Protein engineering pipeline

1. Generate variant sequences from design software
2. Format with variant identifiers for tracking
3. Batch validation of all variants
4. Submit for solubility prediction in parallel
5. Compare results across variant libraries
6. Select optimal candidates for experimental testing

Troubleshooting input issues

Common validation errors

Invalid amino acids:

• Check for non-standard amino acid codes
• Remove ambiguous characters (B, J, O, U, X, Z)
• Verify sequence source and translation accuracy
• Use sequence cleaning tools if needed

Length violations:

• Split very long sequences into domains
• Check for sequence concatenation errors
• Verify translation frame for DNA-derived sequences
• Consider domain-specific solubility analysis

File format issues:

• Ensure proper FASTA header formatting
• Check for encoding problems (use UTF-8)
• Remove empty lines and formatting artifacts
• Validate with text editors or sequence tools

Duplicate sequences:

• Implement unique identifier schemes
• Remove true duplicates from input files
• Use version numbers for variant tracking
• Consider if duplicates are intentional (controls)

Best practices for input preparation

Quality assurance

• Always validate sequences before submission
• Use consistent naming conventions across projects
• Document sequence sources and modifications
• Keep backup copies of original sequence files

Performance optimization

• Batch similar-length sequences together
• Compress large files to reduce upload time
• Use descriptive but concise identifiers
• Consider splitting very large batches into manageable jobs

Result traceability

• Include meaningful identifiers for result interpretation
• Document sequence modifications and sources
• Maintain version control for sequence datasets
• Link predictions back to experimental constructs

Read next