GEO Accession viewer

NCBI > GEO > Accession Display

Not logged in | Login

GEO help: Mouse over screen elements for information.

Sample GSM1382510

Query DataSets for GSM1382510

Status

Public on May 08, 2015

Title

Sample type

SRA

Source name

blood stage

Organism

Plasmodium falciparum

Characteristics

strain: 3D7
developmental stage: 6 hpi

Treatment protocol

We synchronized cultures using multiple 5% sorbitol solution treatments, and expanded cultures to accommodate harvesting of at least 50 mL of culture at each planned time-point (time-course 1: every ~4 hours for 56 hours; time course 2: ~4 hours before and after the ring to trophozoite and trophozoite to schizont morphological stage transitions).

Growth protocol

In each experiment, we cultured a freshly thawed P. falciparum strain 3D7 clone in human RBCs using standard methods. We maintained cultures at 4% hematocrit and supplemented RPMI-HEPES medium with 5% human serum (O+) and 5% Albumax II (Gibco).

Extracted molecule

total RNA

Extraction protocol

We centrifuged 50 mL aliquots of harvested culture at 2400 rpm in a Sorval RT6000B to obtain at least 2 mL of packed RBCs per time-point. For time-course 1, we stored packed RBCs in 15 mL of Buffer RLT (with BME added) at -80°C prior to RNA extraction. For time-course 2, we lysed packed RBCs using a .05% saponin solution, washed liberated parasites using phosphate-buffered saline (pH 7.4), pelleted parasites at 13.2 RPM in a micro-centrifuge, and stored parasites in 1 mL of TRIZOL reagent at -80°C prior to RNA extraction. For time-course 1, we thawed RBC samples stored in Buffer RLT, added one volume of 70% ethanol, and immediately loaded the mixture onto RNeasy Midi columns (Qiagen). For time-course 2, we thawed parasite samples stored in TRIZOL reagent, performed TRIZOL-chloroform extraction, and immediately applied the aqueous layer to RNeasy Mini columns (Qiagen). During both RNeasy Midi and Mini RNA extraction procedures, we performed the optional on-column DNase I digestion for thirty minutes to remove genomic DNA. We stored isolated total RNA aliquots at -80°C with 1 unit/uL RNaseOUT (Invitrogen), and validated RNA quality using an Agilent Bioanalyzer RNA 6000 Pico Kit.
We began library preparation with a second DNase treatment (Ambion TURBO DNase) using 20 units of SUPERase-In (Ambion) and 40 units of RNaseOUT (Invitrogen) to protect RNA from degradation. Each DNase reaction was incubated at 25°C for 30 minutes followed by 1.8X RNAclean SPRI bead purification (Agencourt). Second, we used a Human/Mouse/Rat Ribo-Zero Magnetic Kit (Epicentre) to deplete 18S and 28S rRNA from DNase-treated total RNA. We used 3.5-5 ug of DNase-treated total RNA for all samples except T6, T14, and TT8. In these cases, we used .4 ug, 1 ug, and .7 ug of DNase-treated total RNA, respectively. Furthermore, for T6 we mixed 4 and 8 hpi total RNA 1:1, and for T14 we mixed 12 and 16 hpi total RNA 1:1. Third, we fragmented rRNA-depleted RNA at 85°C for 8 minutes using Mg2+ Fragmentation Buffer (New England Biolabs), followed by 1.8X RNAclean SPRI bead purification (Agencourt). Fourth, we reverse-transcribed the fragmented RNA using SuperScript III (Invitrogen), 200 ng/uL of freshly prepared Actinomycin D (Sigma-Aldrich), 3 ug of 76% AT-biased random hexamers (Integrated DNA Technologies), and a gradually ramping up thermocycler program. Specifically, we set the ramp speed of a PTC-225 DNA Engine Tetrad (MJ Research) to .1°C/second, and used the following program: 5°C, 10°C, 15°C, 20°C, 25°C, 30°C, 35°C for 5 minutes each, 42°C for 30 minutes, 45°C, 50°C, 55°C for 10 minutes each. We cleaned up first strand synthesis (FSS) reactions using both Micro Bio-Spin P-30 Rnase-free columns (Bio-Rad) and 1.8X RNAclean SPRI bead purification (Agencourt). Fourth, we performed second strand synthesis (SSS) using a biased dACG-TP/dU-TP mix (Fermentas), 10 units of E. coli DNA ligase (Invitrogen), 160 units of E. coli DNA polymerase (Invitrogen), and 2 units of E. coli RNase H (Invitrogen), followed by 1.8X AmpureXP SPRI bead purification (Agencourt). Fifth, we used an Illumina series KAPA Library Preparation Kit (Kapa Biosystems) and barcoded Y-adapters developed by the Broad Institute to end repair, A-tail, and ligate adapters to each library. We added adapters in approximately 15-fold excess of library targets, and removed un-ligated adapters and adapter-dimers using 1.0X AmpureXP SPRI bead purification (Agencourt). Sixth, we digested the dUTP-marked second strand at 37°C for 30 minutes, followed by 25°C for 15 minutes, using Uracil-Specific Excision Reagent (USER) enzyme (New England Biolabs). Seventh, we amplified libraries for as few cycles as necessary using a KAPA Real-Time PCR Library Amplification Kit (Kapa Biosystems) and PCR primers developed by the Broad Institute. Each library except for T6, T14, and TT8 required only four PCR cycles, while T14 and TT8 required eight PCR cycles, and T6 required twelve PCR cycles. Following a 2 minute denaturation step at 98°C, we cycled libraries using an ABI 7900 Real-Time PCR machine and the following 2-step program: (1) denaturation at 98°C for 20 seconds, (2) annealing and extension at 55°C for 190 seconds. Finally, we quantified libraries using a KAPA Library Quantification Kit (Kapa Biosystems) and combined barcoded libraries into two pools. We sequenced each pool on an Illumina Hiseq machine (one lane per pool) using 101-bp, paired-end read technology. We prepared the fifteen libraries used in this study in parallel (except for real-time amplification).

Library strategy

ncRNA-Seq

Library source

transcriptomic

Library selection

cDNA

Instrument model

Illumina HiSeq 2000

Description

processed data files: PlasmoDBv10_Expression_FPKMs.txt, Significant_Features.xlsx, LncRNA_Properties.xlsx

Data processing

We trimmed the last base from reads passing Illumina filtering (PF) using Picard release 1.94 (SamToFastq).
We then aligned reads from each sample to the PlasmoDBv10.0 reference genome using TopHat v2.0.9 and the following parameters: -r 300 –mate-std-dev 100 –library-type fr-firststrand –i 70 –I 5000 –read-mismatches 0 –segment-mismatches 0 –max-segment-intron 5000 –max-coverage-intron 5000 –b2-very-sensitive –read-gap-length 0 –read-edit-dist 0 –read-realign-edit-dist 0 –max-deletion-length 0 –max-insertion-length 0 –max-multihits 2 –no-mixed –no-discordant. These parameters specified an average fragment size of 300 bp with a standard deviation of 100 bp, strand-specific reads prepared using the dUTP method, and an expected intron size of 70-5,000 bp. Also, to allow only perfect read pair mapping in the expected orientation, and to report only two alignments for non-unique read pairs.
To remove non-unique reads, we then used SAMtools v0.1.19 (view –q 10) to remove read pairs with a mapping quality of 10 or less, meaning at least a 10% chance that the read pair truly came from somewhere else in the genome.
We used Cufflinks v2.1.1 to assemble aligned reads from each sample into transfrags using the following parameters: –library-type fr-firststrand –max-intron-length 5000 –overlap-radius 1 –min-isoform-fraction .25 –pre-mrna-fraction .25 –min-frags-per-transfrag 50 –trim-3-dropoff-frac .2 –frag-bias-correct. These parameters specified strand-specific reads prepared using the dUTP method and a maximum intron size of 5,000 bp. Also, to not merge transfrags, to filter minor isoforms less than 25% as abundant as the major isoform, to ignore intronic alignments if not as abundant as specified, to require at least fifty aligned reads per assembled transfrag, to trim the 3’ end of assembled transfrags at 20% of average coverage, and to run the built-in bias correction algorithm prior to estimating transcript abundance. We also specified a set of 374 short (<300 bp) and/or structural RNAs (such as tRNAs, rRNAs, and snoRNAs) to exclude from assembly and abundance estimation.
We used Cuffmerge to parsimoniously merge assembled transfrags from each sample into a final transcriptome assembly, incorporating PlasmoDBv10.0 annotated transcript models (RABT). In sum, annotation-assisted Cufflinks RABT assembly predicted 9434 transcripts, including 660 un-annotated intergenic transcripts (647 unique loci) and 474 antisense transcripts (467 unique loci). The 467 unique antisense loci overlapped 462 annotated genes in an approximately 1:1 ratio. This encompassed transcription of at least 73% of the P. falciparum genome, a 13% increase compared to annotation alone, and included the prediction of high-confidence antisense transcripts from 8% of annotated genes.
We used TransDecoder release 2013-11-17 to predict coding regions in the Cufflinks RABT transcriptome. TransDecoder identifies regions in spliced transcript models that likely encode peptides greater than 100 amino acids based on (1) a match to a Pfam domain above the noise threshold or (2) a log-likelihood score of a Markov model for coding DNA that is greater than zero and greatest when calculated in the predicted open reading frame. Using this approach, we predicted coding regions in 11 out of 660 intergenic transcripts (1.7%) and 7 out of 474 antisense transcripts (1.4%), versus 5213 out of 5229 assembled protein-coding transcripts at least 100 amino acids long (99.7%)
To assess differential expression and regulation of both the PlasmoDBv10.0 and the Cufflinks RABT transcriptomes, we used Cuffdiff v2.1.1 and the following parameters: –library-type fr-firststrand –time-series –min-reps-for-js-test 1. These parameters specified strand-specific reads prepared using the dUTP method, time-course samples, and to not require all conditions to be replicated. We also instructed Cuffdiff to compare samples from the 56-hour time-course harvested approximately 8 hours apart (T6 vs. T14, T14 vs. T24, T20 vs. T28, T24 vs. T32, T28 vs. T36, T32 vs. T40, T36 vs. T44, T40 vs. T48, T48 vs. TT8) and to estimate biological variation across the blood stage using the four replicated conditions. Specifically, we considered the correlation matrix between sample profiles and paired R with T14 and T20, ET with T28, LT with T32, and S with T40. We then supplied these replicated conditions to Cuffdiff as samples in addition to specifying the desired sample comparisons described above. Finally, we used the Benjamini and Hochberg method and a FDR cut-off of .05 to threshold significance. After removing chimeric results, our analysis of the Cufflinks RABT transcriptome predicted 3815 differentially expressed genes, 127 alternative splicing events, and 81 cases of alternative promoter usage. This included 354 significant intergenic lncRNA loci (out of 647) and 69 significant antisense loci (out of 467). [See PlasmoDBv10_Expression_FPKMs.txt, Significant_Features.xlsx, LncRNA_Properties.xlsx].
Genome_build: PlasmoDB 10.0
Supplementary_files_format_and_content: We generated Gene Transfer Format (gtf) files using either Cufflinks or TransDecoder. Fields are as specified by UCSC; The abundance measurement tab-delimited text file contains FPKM values estimated by Cuffdiff for the 5,777 PlasmoDBv10.0 annotated transcripts. Masked transcripts (short or structural RNAs) have FPKMs of zero in every sample; The .xlsx matrix tables list properties as specified in the headers of each worksheet for the 3815 differentially expressed genes, 127 alternative splicing events, 81 cases of alternative promoter usage, 660 intergenic lncRNAs, and 474 antisense transcripts predicted in the Cufflinks RABT transcriptome. We assembled transcripts using Cufflinks/Cuffmerge RABT and computed significance using Cuffdiff.
Supplementary_files_format_and_content: Processed data files [PlasmoDBv10_Expression_FPKMs.txt, Significant_Features.xlsx, LncRNA_Properties.xlsx] are linked as supplementary files on the Series record.

Submission date

May 08, 2014

Last update date

May 15, 2019

Contact name

Kate Mariel Broadbent

E-mail(s)

kbroadb@fas.harvard.edu

Organization name

Harvard University

Department

Systems Biology