-- 
-- Creating entities for Deep Sequencing Distribution 
--

-- @author Manuel Kohler

set type = POLYCOMB2010
register-type EXPERIMENT ${type}

register-type SAMPLE ILLUMINA_FLOW_CELL listable=true show-container=false show-parents=false description="Flow Cell containing eight flow lanes"
register-type SAMPLE ILLUMINA_FLOW_LANE listable=true show-container=true show-parents=true description="Flow Lane is parented by one ILLUMINA_SEQUENCING"
register-type SAMPLE ILLUMINA_SEQUENCING listable=true show-container=false show-parents=false description="Biological Sample"

set type = FASTQ_GZ
register-type DATA_SET ${type} description="Gzipped FASTQ files produced by CASAVA 1.8+"

set type = ILLUMINA_GA_OUTPUT
register-type DATA_SET ${type} description="Illumina GA Output"

set type = ILLUMINA_HISEQ_OUTPUT description="Illumina HiSeq Output"
register-type DATA_SET ${type}

set type = MACS_OUTPUT
register-type DATA_SET ${type} description="MACS Peak Caller output"

set type = TSV
register-type DATA_SET ${type} description="All kind of Tab Separated Value files"

set type = WIGGLE
register-type DATA_SET ${type} description="Visualization Format of ChIP-Seq data for UCSC browser"

set type = BED
register-type DATA_SET ${type} description="Visualization Format of ChIP-Seq data for UCSC browser"

set type = AFFILIATION
register-property-type ${type} with data type CONTROLLEDVOCABULARY(AFFILIATION) "description=Where data will be shipped to" "label=Affiliation" 
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = ALIGNMENT
register-type DATA_SET ${type} description="Alignment software output, e.g.  SAM/BAM"
assign-to DATA_SET:${type} AFFILIATION

set type = AGILENT_KIT
register-property-type ${type} with data type CONTROLLEDVOCABULARY(AGILENT_KIT) "description=Kit used for QC" "label=Agilent Kit" 
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = ANALYSIS_FINISHED
register-property-type ${type} with data type TIMESTAMP "description=Post processing of  the complete Flow Cell is finished and the fastq/SRF files have been created" "label=Analysis finshed" 
assign-to SAMPLE:ILLUMINA_FLOW_CELL ${type}

set type = BIOLOGICAL_SAMPLE_ARRIVED
register-property-type ${type} with data type TIMESTAMP "description=The date when the biological sample arrived at lab" "label=Arrival Date of Biological Sample" 
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = BARCODE
register-property-type ${type} with data type CONTROLLEDVOCABULARY(BARCODES) "description=Barcodes used for multiplexing" "label=Barcode"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = BARCODE_LENGTH
register-property-type ${type} with data type INTEGER "description=Length of the barcode. Will be used as a parameter for the bareback image analysis: Large Scale Loss of Data in Low-Diversity Illumina Sequencing Libraries Can Be recovered by Deferred Cluster Calling Felix Krueger1, Simon R. Andrews1, Cameron S. Osborne2*,Plos One, 2011" "label=Barcode Length (+ recognition site)"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = BAREBACKED
register-property-type ${type} with data type BOOLEAN "description=Has the bareback algorithm been applied? Krueger F, Andrews SR, Osborne CS (2011) Large Scale Loss of Data in Low-Diversity Illumina Sequencing Libraries Can Be Recovered by Deferred Cluster Calling. PLoS ONE 6(1): e16607. doi:10.1371/journal.pone.0016607" "label=Barebacked?"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = CLUSTER_GENERATION_KIT_VERSION
register-property-type ${type} with data type CONTROLLEDVOCABULARY(CLUSTER_GENERATION_KIT_VERSION) "description=Cluster Station Kit Version" "label=CS Generation Kit Version"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = CASAVA_VERSION
register-property-type ${type} with data type CONTROLLEDVOCABULARY(CASAVA_VERSION) "description=Illumina's post analyzing software" "label=Casava Version"
assign-to DATA_SET:ILLUMINA_GA_OUTPUT ${type}
assign-to DATA_SET:ILLUMINA_HISEQ_OUTPUT ${type}

set type = CLUSTER_STATION
register-property-type ${type} with data type CONTROLLEDVOCABULARY(CLUSTER_STATION) "description=Used Cluster Station" "label=Cluster Station"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = CONCENTRATION_ORIGINAL
register-property-type ${type} with data type REAL "description=Concentration of the original sample for Illumina Sequencing in nano grams per micro liter" "label=Concentration (original) [ng/μl]"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = CONCENTRATION_PREPARED
register-property-type ${type} with data type REAL "description=Concentration of the prepared sample for Illumina Sequencing in nano grams per micro liter" "label=Concentration (prepared) [ng/μl]"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = CONCENTRATION_FLOW_LANE
register-property-type ${type} with data type REAL "description=Concentration of library loaded in flow lane" "label=Concentration in flow lane [pM]"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = CYCLES_REQUESTED_BY_CUSTOMER
register-property-type ${type} with data type CONTROLLEDVOCABULARY(CYCLES) "description=Read length" "label=Cycles"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}
assign-to SAMPLE:ILLUMINA_FLOW_CELL ${type}

set type = DNA_CONCENTRATION_OF_LIBRARY
register-property-type ${type} with data type INTEGER "description=" "label=DNA concentration of library (nM)"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = CONCENTRATION_TOTAL
register-property-type ${type} with data type INTEGER "description=Total amount of genetic material" "label=DNA concentration of library (ng/µl)"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = DATA_TRANSFERRED
register-property-type ${type} with data type TIMESTAMP "description=Date of successful data transfer to the customer" "label=Data transferred on"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = CONTACT_PERSON_EMAIL
register-property-type ${type} with data type VARCHAR "description=Email of person to contact about the sample, Used for the tracking system" "label=Email of Contact Person"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type} mandatory

set type = CONTACT_PERSON_NAME
register-property-type ${type} with data type VARCHAR "description=Name of person to contact about the sample" "label=Email of Contact Person"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type} mandatory

set type = EXTERNAL_SAMPLE_NAME
register-property-type ${type} with data type VARCHAR "description=Original Sample Name" "label=External Sample Name"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type} mandatory

set type = FRAGMENT_SIZE_PREPARED_ILLUMINA
register-property-type ${type} with data type INTEGER "description=Fragment size of the library of a prepared Illumina Sequencing sample in either bases or base pairs" "label=Fragment Size (prepared) [base (pairs)]"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = INVOICE
register-property-type ${type} with data type BOOLEAN "description=Check box if invoice has been sent" "label=Invoice sent?"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = LOT
register-property-type ${type} with data type INTEGER "description=Illumina Kit Lot" "label=Kit Lot #"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = KIT_ARRIVED
register-property-type ${type} with data type TIMESTAMP "description=Date when Illumina Kit arrived" "label=Kit arrived"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = LIBRARY_PROCESSING_POSSIBLE
register-property-type ${type} with data type BOOLEAN "description=" "label=Library Processing possible"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = KIT
register-property-type ${type} with data type CONTROLLEDVOCABULARY(KIT) "description=Type of preparation kit" "label=Type of preparation kit"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = LIBRARY_PROCESSING_SUCCESSFUL
register-property-type ${type} with data type BOOLEAN "description=" "label=Library processing successful"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = MACS_VERSION
register-property-type ${type} with data type CONTROLLEDVOCABULARY(MACS_VERSION) "description=Used MACS version (peak caller) http://liulab.dfci.harvard.edu/MACS/" "label=MACS Version"
assign-to DATA_SET:MACS_OUTPUT ${type}

set type = MAPPED_READS
register-property-type ${type} with data type INTEGER "description=Reads successfully aligned to reference genome" "label=# of mapped reads"
assign-to  DATA_SET:ALIGNMENT ${type}

set type = NANO_DROP
register-property-type ${type} with data type CONTROLLEDVOCABULARY(NANO_DROP) "description=" "label=Nano Drop"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = NCBI_ORGANISM_TAXONOMY
register-property-type ${type} with data type CONTROLLEDVOCABULARY(NCBI_TAXONOMY) "description=Standard NCBI Taxonomy (http://www.ncbi.nlm.nih.gov/taxonomy)" "label=Organism (NCBI Taxonomy)"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type} mandatory

set type = NOTES
register-property-type ${type} with data type MULTILINE_VARCHAR "description=Free text" "label=Notes"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}
assign-to SAMPLE:ILLUMINA_FLOW_CELL ${type}

set type = NOTES_CUSTOMER
register-property-type ${type} with data type MULTILINE_VARCHAR "description=Free text" "label=Customer notes"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = NUMBER_OF_ATTACHMENTS
register-property-type ${type} with data type INTEGER "description=" "label=# of attachments"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = OTRS
register-property-type ${type} with data type INTEGER "description=OTRS Reference Number" "label=OTRS ID #"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = PRINCIPAL_INVESTIGATOR_EMAIL
register-property-type ${type} with data type VARCHAR "description=Email of Principal Investigator responsible for the sample. Used for the tracking system" "label=Email of Principal Investigator"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type} mandatory

set type = PRINCIPAL_INVESTIGATOR_NAME
register-property-type ${type} with data type VARCHAR "description=Name of Principal Investigator responsible for the sample." "label=Name of Principal Investigator"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type} mandatory

set type = REQUIRED_LANES
register-property-type ${type} with data type CONTROLLEDVOCABULARY(REQUIRED_LANES) "description=" "label=Required Lanes"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = SAMPLE_CHARACTERISTICS
register-property-type ${type} with data type MULTILINE_VARCHAR "description=Describe all available characteristics of the biological source, including factors not necessarily under investigation. Provide in 'Tag: Value' format, where 'Tag' is a type of characteristic (e.g. 'gender', 'strain', 'tissue', 'developmental stage', 'tumor stage', etc), and 'Value' is the value for each tag (e.g. 'female', '129SV', 'brain', 'embryo', etc). Include as many characteristics fields as necessary to thoroughly describe your Samples." "label=Sample characteristics"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = SAMPLE_DATA_PROCESSING
register-property-type ${type} with data type MULTILINE_VARCHAR "description=Provide details of how data were generated and calculated. For example, what software was used, how and to what were the reads aligned, what filtering parameters were applied, how were peaks calculated, etc. Include a separate 'data processing' attribute for each file type described." "label=Sample Data Processing"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = SAMPLE_EXTRACT_PROTOCOL
register-property-type ${type} with data type MULTILINE_VARCHAR "description=Describe the protocol used to isolate the extract material. Describe the library construction protocol, ie, the protocols used to extract and prepare the material to be sequenced. You can include as much text as you need to thoroughly describe the protocol; it is strongly recommended that complete protocol descriptions are provided within your submission." "label=Sample Extract Protocol"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = SAMPLE_KIND
register-property-type ${type} with data type CONTROLLEDVOCABULARY(SAMPLE_TYPE) "description=Kind of sample delivered by the customer (and suitable for a certain sequencing application)" "label=Sample Kind"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type} mandatory

set type = SAMPLE_LIBRARY_SELECTION
register-property-type ${type} with data type CONTROLLEDVOCABULARY(SAMPLE_LIBRARY_SELECTION) "description=Describes whether any method was used to select and/or enrich the material being sequenced." "label=Sample Library Selection"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = SAMPLE_LIBRARY_SOURCE
register-property-type ${type} with data type CONTROLLEDVOCABULARY(SAMPLE_LIBRARY_SOURCE_VOC) "description=Type of source material that is being sequenced." "label=Sample Library Source"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = SAMPLE_LIBRARY_STRATEGY
register-property-type ${type} with data type CONTROLLEDVOCABULARY(SAMPLE_LIBRARY_STRATEGY) "description=Sequencing technique for this library." "label=Sample Library Strategy"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = SAMPLE_MOLECULE
register-property-type ${type} with data type CONTROLLEDVOCABULARY(SAMPLE_MOLECULE) "description=Specify the type of molecule that was extracted from the biological material." "label=Sample molecule"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = SAMPLE_SOURCE_NAME
register-property-type ${type} with data type VARCHAR "description=Briefly identify the biological material and the experimental variable(s), e.g., vastus lateralis muscle, exercised, 60 min." "label=Sample Source Name"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = SAMTOOLS_FLAGSTAT
register-property-type ${type} with data type MULTILINE_VARCHAR "description=Output of the samtools flagstat command" "label=Samtools Flagstat Output"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = SBS_SEQUENCING_KIT_VERSION
register-property-type ${type} with data type CONTROLLEDVOCABULARY(SBS_SEQUENCING_KIT_VERSION) "description=SBS Sequencing Kit Version" "label=SBS Sequencing Kit Version"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = SCS_PROTOCOL_VERSION
register-property-type ${type} with data type CONTROLLEDVOCABULARY(SCS_PROTOCOL_VERSION) "description=Illumina protocol version" "label=SCS Protocol Version"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = SCS_SOFTWARE_VERSION
register-property-type ${type} with data type CONTROLLEDVOCABULARY(SCS_SOFTWARE_VERSION) "description=Version of the SCS Software on the GA Control PC" "label=SCS Software Version"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = SEQUENCER
register-property-type ${type} with data type CONTROLLEDVOCABULARY(SEQUENCER) "description=Which sequencer has been used?" "label=Sequencer"
assign-to SAMPLE:ILLUMINA_FLOW_CELL ${type}

set type = SEQUENCING_APPLICATION
register-property-type ${type} with data type CONTROLLEDVOCABULARY(SEQUENCING_APPLICATION) "description=What kind of sequencing application the sample has been prepared for" "label=Sequencing Application"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type} mandatory

set type = STARTING_AMOUNT_OF_SAMPLE_IN_NG
register-property-type ${type} with data type REAL "description=" "label=Starting amount of sample (ng)"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = TOTAL_READS
register-property-type ${type} with data type REAL "description=Total reads which have been tried to map" "label=Total reads"
assign-to DATA_SET:ALIGNMENT ${type}

set type = CLUSTERING_DATE
register-property-type ${type} with data type TIMESTAMP "description=Date of Clustering Process" "label=Clustering Date"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}

set type = CLUSTERING_PROTOCOL_VERSION
register-property-type ${type} with data type VARCHAR "description=" "label=Cluster Station Protocol Version"
assign-to SAMPLE:ILLUMINA_SEQUENCING ${type}