Data Employed in the Construction of a Composite Protein Database for Proteogenomic Analyses of Cephalopods Salivary Apparatus

Database A (Database_A_19087_sequences.fasta) included in Dataset_1 is the same comprehensive database deposited by Fingerhut et al. (2018) under the identifier PXD010298 at ProteomeXchange via PRIDE (OVulgarisMQ_20172206.fasta) [2]. Briefly, this database was provided as a FASTA file (5.9 MB) accounting for a total of 19,087 protein sequences. This database included 18,536 protein sequences (called “known” by the authors) predicted by TransDecoder, embedded in Trinity v3.0.1, and 354 protein sequences (called “novel” by the authors) obtained with the six-frame translation tool and validated at the proteomic level (using the script “sixframe.rb” available as part of the Protk toolkit [4]). Both, “known” and “novel” proteins were obtained from the transcriptome of the PSGs of O. vulgaris [2]. Moreover, the database included 197 protein sequences of cephalopods retrieved UniProt, inferred from the O. vulgaris saliva proteome. The detailed description of how this database was constructed can be found in Fingerhut et al. (2018) [2].

Database B (Database_B_16990_sequences.fasta) included in Dataset_1 contains one of the most comprehensive collections of non-redundant AMPs, which comprises 16,990 AMPs carefully gathered by Aguilera-Mendoza et al. (2015) from 25 AMPs databases [3]. Some of these proteins came from the UniProt database. Details related to this AMPs database construction can be found in the original article [3]. In order to perform MaxQuant analyses, the names of these sequences were edited through the removal of all the characters located after the first space detected.

Database C (Database_C_2427_sequences.fasta), provided in Dataset_1, contains original PSGs proteomes from three O. vulgaris specimens comprising 2427 protein sequences identified with Proteome Discoverer software v2.2.0.388 (Thermo Scientific, Waltham, MA, USA) against the Orbitrap raw data deposited at the Mendeley Data repository (http://dx.doi.org/10.17632/csc8shzkwc.1; http://dx.doi.org/10.17632/787g95ppwv.1; http://dx.doi.org/10.17632/8fhx775zdf.1; http://dx.doi.org/10.17632/d6wxyt22kx.1; http://dx.doi.org/10.17632/zbtkf2nsvh.1 and http://dx.doi.org/10.17632/df4cbg73tx.1) [1]. All these protein sequences have a name composed of the UniProt accession followed by “_PD”, indicating that those sequences came from the Proteome Discoverer analysis (e.g., sp|P18499_PD). Details about sample preparation for LC–MS/MS analysis and further protein identification using Proteome Discoverer for the construction of this database can be found in the two steps described below (STEP 1 and STEP 2).

Briefly, protein samples from PSGs comprising three biological replicates of O. vulgaris caught in the eastern Atlantic (Portuguese waters) were processed in duplicate following two distinct protocols (i.e., total of six protein samples for each protocol) as follows: filter-aided sample preparation (FASP) [5] and in-solution digestion (ISD) using RapiGest SF Surfactant according to the manufacturer’s specifications (Waters Corporation, Milford, MA, USA). Protein samples prepared according to FASP and ISD protocols were processed using a nano LC–MS/MS, composed of an Ultimate 3000 liquid chromatography system coupled to a Q-Exactive Hybrid Quadrupole-Orbitrap mass spectrometer (Thermo Scientific, Bremen, Germany).

The raw data from LC–MS/MS analysis corresponding to 12 Orbitrap files of PSGs from O. vulgaris were processed using Proteome Discoverer software v2.2.0.388 (Thermo-Fisher, Waltham, MA, USA) and searched against the UniProt database for the Metazoan taxonomic selection (https://www.uniprot.org/taxonomy/33208; 2018_07 release) [6]. The Sequest HT search engine was used for protein identification. The ion mass tolerance was 10 ppm for precursor ions and 0.02 Da for fragment ions. The maximum number of missing cleavage sites allowed was set to 2. Cysteine carbamidomethylation was defined as a constant modification. Methionine oxidation and protein N-terminus acetylation were defined as variable modifications. Peptide confidence was set to high. The processing node Percolator was enabled with the following settings: maximum delta Cn 0.05, decoy database search target FDR 1%, validation was based on q-value. The output files from this analysis were provided in Dataset_2.

Briefly, the workflow to construct Databases D (Database_D_84778_sequences.fasta) and E (Database_E_5106635_sequences.fasta), both included in Dataset_1, consisted of the following steps: (i) a first analysis of all the resulting contigs (272,704 contigs) from the de novo assembly of the cephalopods’ PSGs transcriptomes with the TransDecoder v5.5.0 tool [7], predicting a total of 84,778 protein sequences grouped in Database D; and (ii) an analysis with the six-frame translation tool (“sixframe.rb” script [4]) of those contigs that did not produce protein-coding sequences when analysed by the TransDecoder v5.5.0 tool (i.e., 187,926 contigs discarded by TransDecoder when considering default parameters and not included in Database D), whose results compose Database E (5,106,635 protein sequences). Details about the construction of these databases can be found in Table 2 and in the three steps described below (STEP 1 to STEP 3).

In order to obtain the available data from transcriptomes of the cephalopods’ PSGs, a search at the Sequence Read Archive (SRA) of National Center for Biotechnology Information (NCBI) was first performed. This search was based on the following criteria: (i) the term “Cephalopoda” was considered [8], all returned results were selected, sent to “Run selector” tool and posteriorly downloaded in a Tab delimited format (SraRunTable.txt); (ii) the SraRunTable file was then inspected to select all the SRA run accessions (SRR#) associated with any of these terms, namely “poison gland”, “posterior venom gland” and “posterior salivary gland/glands”, recovering reads from seven transcriptomes (SRR6349992, SRR2047107, SRR3105321, SRR3105558, SRR7130741, SRR5204441 and SRR5204442). Then, to get additional information, a search with the same term “Cephalopoda” was performed at the Sequence Set Browser from NCBI [9], which after being filtered by “posterior venom gland” retrieved 10 results (i.e., BioProject accessions: PRJNA188569, PRJNA188570, PRJNA188571, PRJNA188572, PRJNA188573, PRJNA188575, PRJNA188576, PRJNA188577, PRJNA188658 and PRJNA188659). These BioProject accessions were searched at the SRA of NCBI in order to obtain their corresponding SRR#, which were compiled together with the previous ones, resulting in a total of 17 transcriptomes from cephalopods’ PSGs. From these 17 transcriptomes, SRR7130741 was discarded in order to avoid redundancy in further MaxQuant analyses since it corresponded to the O. vulgaris PSGs transcripts [2] already included within Database A. Then, the FASTQ files of each one of the remaining 16 transcriptomes (SRR6349992, SRR2047107, SRR3105321, SRR3105558, SRR5204441, SRR5204442, SRR680047, SRR725938, SRR725597, SRR725937, SRR725936, SRR684223, SRR725779, SRR684167, SRR725935 and SRR725780) [10] were downloaded from the European Nucleotide Archive [11] for further assembly. The FASTQ files of each transcriptome were imported to CLC Genomics Workbench v11.0.1 [12] for adapter trimming and further de novo assembly. The reads present within the FASTQ files were first trimmed considering a list of adapters and possible contaminants, provided within Dataset_3 (Table_S1.xlsx), in case of being paired reads, or by removing terminal nucleotides of each read (i.e., 10 nucleotides from the 5′ and 3′ ends), in case of being single reads, and finally assembled using default parameters (Table 2). All the resulting assemblies were provided as Dataset_3.

All the contigs resulting from the CLC de novo assemblies (total of 272,704 contigs) provided in Dataset_3 (272704_contigs_from_16_cephalopods_PSGs_transcriptome_assemblies.fasta) were then conducted to the TransDecoder v5.5.0 tool [7], using default parameters, in order to predict protein-coding regions in transcripts. The TransDecoder v5.5.0 tool extracted open reading frames (ORFs) that were at least 100 amino acids (AA) long regardless of coding potential and then predicted which of them were likely to be coding (Table 2). The resulting proteins predicted by TransDecoder from the de novo assembly of 16 transcriptomes of PSGs from cephalopods were provided within Dataset_4. All the files provided in Dataset_4 were compiled into a single FASTA file (total of 84,778 protein sequences) using Geneious v11.1.2 [13], and the sequence names were edited with the following format: “SRR#_c#_g#”, where “#” is a number, “c” means contig and “g” means gene, thus creating Database D provided in Dataset_1 (Database_D_84778_sequences.fasta).

In order to prevent the loss of relevant peptides (such as antimicrobial peptides) shorter than the TransDecoder minimum protein length threshold of 100 AA (i.e., default parameters), the contigs not included in Database D (i.e., 187,926 contigs) were analysed by the six-frame translation tool (“sixframe.rb” script [4]). The details of this approach can be found below and in Table 2.

All the files mentioned in the paragraph below are available in Dataset_5.

First, two tables in a csv (comma-separated values) format were prepared. One of the tables was entitled as “cases.csv”, consisting of one column with the header “Name” listing all the SRR# corresponding to the sequences present within Database D (i.e., 84,778 contigs in the format, e.g., SRR684167_c1). The other table was named as “transcripts.csv” and contains two columns, “Name” and “Sequence”, with all the de novo assembled transcript names (SRR#_c#) and corresponding nucleotide sequences deposited in Dataset_3 (i.e., 272,704 contigs). Then, a database (DB.db) was created using the DB Browser for SQLite v3.11.2 [14] and the two tables in csv format were imported. After executing the SQL command (SQL_command.txt), the resulting contigs not included in Database D were exported as a csv table (187926_contigs_not_included_in_Database_D.csv) and posteriorly converted to a FASTA file (187926_contigs_not_included_in_Database_D.fasta) using Geneious v11.1.2 [13]. Furthermore, the “187926_contigs_not_included_in_Database_D.fasta” file was analysed by the six-frame translation tool (“sixframe.rb” script [4]), using default parameters with the exception of the “--min-len” parameter that was set to 10. In this way, 5,106,635 protein sequences with sequence length greater than or equal to 10 AA (six-frame_translation_of_187926_contigs_not_included_in_Database_D.fasta) were obtained. Finally, the names of the sequences within the “six-frame_translation_of_187926_contigs_not_included_in_Database_D.fasta” file were edited by removing everything after the space in order to obtain the final database for MaxQuant analysis, and it was deposited in Dataset_1 (Database_E_5106635_sequences.fasta).

Database F contains 720,910 protein sequences provided in Dataset_1 (Database_F_720910_sequences.fasta) corresponding to the ORFs from O. vulgaris PSGs transcriptomes not included in Database A. Details for the construction of this database can be found below.

First, the files “Database_A_19087_sequences.fasta” (provided in Dataset_1) and “GGNR01.1.fsa_nt.gz” (46,490 contigs deposited under accession PRJNA464423) were opened with Geneious v11.1.2 [13], and the sequence names within both files were edited with the following format: “TRINITY_DN0_c0_g1_i1”. Then, two csv files, each one from the previously edited files, were saved and made available in Dataset_6.

Hereafter, all the mentioned files are available in Dataset_6. Specifically, for the preparation of these csv files, we followed two steps: (i) from the “Database_A_19087_sequences.fasta” file, all the names with the “TRINITY” term included were selected and saved as a list, making a total of 18,890 sequence names (cases.csv: consisting of one column with the header “Name”); and (ii) from the “GGNR01.1.fsa_nt.gz” file, a list with all the nucleotide sequences was saved (transcripts.csv: consisting of two columns with the headers “Name” and “Sequence”). Thus, there was a match between the header “Name” of these two saved files. Then, a database (DB1.db) was created using the DB Browser for SQLite v3.11.2 [14] and the two lists in csv format were imported. After executing the SQL command (SQL_command1.txt), the resulting contigs not included in Database A were exported as a csv table (31661_contigs_not_included_in_Database_A.csv) and posteriorly converted to a FASTA file (31661_contigs_not_included_in_Database_A.fasta) using Geneious v11.1.2. The “31661_contigs_not_included_in_Database_A.fasta” file was analysed by the six-frame translation tool (sixframe.rb [4]), using default parameters with the exception of the “--min-len” parameter that was set to 10. In this way, 720,910 protein sequences with sequence length greater than or equal to 10 AA (six-frame_translation_of_31661_contigs_not_included_in_Database_A.fasta) were obtained. Finally, the names of the sequences within the “six-frame_translation_of_31661_contigs_not_included_in_Database_A.fasta” file were edited by removing everything after the space in order to obtain the final database for MaxQuant analysis, and it was deposited in Dataset_1 (Database_F_720910_sequences.fasta).