P. crispa was collected in areas near the Arctowski Polish Station Region, Admiralty Bay, King George Island (61°50′−62°15′ S and 57°30′−59°00′W), Antartic. The collection was carried out in the Antarctic summer, in January of 2014 Austral summer, with temperature ranging from 0.5 to 2.0°C. The samples were maintained in RNAlater® (Sigma-Aldrich, USA) until RNA extraction.

Total RNA was extracted from three pools of samples using an RNAqueous®-Micro Total RNA Isolation Kit (Thermo Fisher Scientific Inc., USA) according to the manufacturer's instructions. The RNA-Seq library was prepared using random primers. The transcriptome was sequenced by Macrogen Service using the Solexa-Illumina HiSeq 2000 next-generation sequencing platform device according to the manufacturer's instructions. A paired-end reads with a read size of ~100 bp separated by insert size of 300 bp was employed.

The BioProject ID of our data is PRJNA329112, and the BioSample accession number is SAMN05392062. All raw reads were deposited into the Sequencing Read Archive (SRA) of NCBI with accession number SRR5754271. This Transcriptome Shotgun Assembly project has been deposited at DDBJ/EMBL/GenBank under the accession GFTS00000000.

Raw reads from data sets were filtered to remove the adapter sequences, and low quality reads with Fastx-toolkit (quality cut-off = 30) (Gordon and Hannon, 2010) and Trimmomatic v 0.36 using default parameters (Bolger et al., 2014). Next, we used Trinity version r2014-07-17 (Ekblom and Galindo, 2011) as Bruijn graph assembler with 25 kmer size. Due to the sequencing of a complex sample extracted from the Antarctic soil, we expected some amount of bacterial and fungal contamination. Therefore, in order to remove such contaminants, we performed a tblastx with default parameters (Altschul et al., 1990) searching against all of the NCBI nucleotide non-redundant database and recovered all contigs in which the best blast hit occurred with algae and plant sequences. Next, we used Bowtie2 (Langmead and Salzberg, 2012) with default parameters to recover only the reads that mapped against those P. crispa contigs.

The assembled and recovered contigs were searched against the NCBI protein non-redundant database using the BLASTX algorithm; the E-value cut-off was set at 1e-10. Genes were tentatively identified based on the best hits against known sequences. Blast2GO (Conesa and Götz, 2008) was used for mapping and annotation, associating Gene Ontology (GO) terms and predicting their function. Assembled and annotated transcriptome is publicly available on Figshare at: https://figshare.com/s/a60ae8a0445d547b9360.

After a stringent filtering process, the processed reads were assembled into 17,201 contigs. Statistics of the assembly are summarized in Table 1. CD-HIT-EST version 4.6.8, 2017-06-21 (Fu et al., 2012) was used for clustering of assembled transcripts with the default parameters with sequence identity threshold set to 95%, in order to indicate the number of unigenes. The clustering reduced the number of transcripts marginally by 3.5%. Our analysis showed that 93% of the unigenes are represented by only one isoform. The metrics of P. crispa were compared with others transcriptomes of the organism from the Trebouxiophyceae class, including Chlorella minutissima (Yu et al., 2016), Trebouxia gelatinosa (Carniel et al., 2016), Coccomyxa subellipsoidea (Peng et al., 2016), Chlorella sorokiniana (Li et al., 2016), Botryococcus braunii (Xu et al., 2015), and was found to have the lowest number of total reads. In relation to the number of contigs, P. crispa is in an intermediary position, with C. sorokiniana being the organism with the highest number of contigs (63,811) and C. subellipsoidea having the lowest number of contigs (9,409). More information on the metrics of transcriptomes is given in Supplementary Table S1.

The search of these contigs against the NCBI protein non-redundant database with BLASTX demonstrates that 8,980 (52.19%) sequences had at least one hit. The mapping of the sequences against the GO database retrieved 7,009 sequences mapped, and all assigned GO terms were classified into three main categories: cellular component, molecular function, and biological process. The distribution of sequences mapped to the three different categories and the top 15 GO terms are represented in Figures 1A,B, respectively.

Thus, taking the comparative numbers of contigs, mean length, the clustering process and sequencing depth and coverage into account its likely that our assembly comprises a representative number of transcripts but which were partially reconstructed. Moreover, it is important to note that although we have used stringent blast parameters to remove contaminations, some contigs can still come from contamination. Therefore, experimental validation to confirm P. crispa origin of some contigs is warranted.

The reading quality of the data of this transcriptomic analysis was evaluated through FastQC software (Babraham Bioinformatics) [RRID:SCR_014583]. The paired-end reads results were merged using MultiQC (http://multiqc.info) and are shown in Supplementary Figure S1. Per base quality phred scores range from 32.78 to 40.06, indicating base call accuracies of >99.9% (Supplementary Figure S1A). Per sequence quality shows that 99.62% of reads had a mean phred score of 30 or above (Supplementary Figure S1B) and per base N content was low, with a maximum value 0.18% (Supplementary Figure S1C).

Through the data of this transcriptome, it is possible to perform searches for genes, aiming the heterologous expression of proteins with biotechnological potential, such as antifreeze proteins, which act to inhibit freezing of intracellular fluids (Nath et al., 2013), heat-shock proteins that play an important role in maintain biological activities in algae present in these acclimatization process (Li and Brawley, 2004) and mycosporine-like amino acids responsive to high incidence of ultraviolet radiation (Kováčik and Pereira, 2001). Proteomics approaches may also be employed, aiming at the confirmation of gene expression at the translational level.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.