ELI What happened? Where's Xena? Him and his main muscle threatened to kill me unless she went with him. Gabrielle looks at tracks. ELI Let's go. Ares appears and watches them. Xena looks around. Xena is in manacles. KAL Bring back memories? XENA I don't know you. KAL You should. I was once on my way to becoming the greatest war god of them all. XENA What do you want with me? Last one I sent after it is probably toast. Purist monk in his order, too. XENA I don't understand. What is it that you want me to get?
ARES Good question. Ares stands behind Kal. He moves over to Xena. What's a nice girl like you doing in manacles like those? Ares makes them fall off. XENA Thank you. ARES Not a problem. Kal draws a sword and Ares gets an axe.
Ares and Kal hit each other with their weapons. The force sends Xena flying back. Kal has an ax in his neck and Ares has a sword in his gut. ARES Uh-huh. They move away and take the weapons out of themselves. They each draw their own swords and attack each other. Lightning flashes. Xena backs away. She sees a drawing of the chakram on the wall. She touches it. She leaves out a window. Ares and Kal continue fighting. The group watches the flashing lightning.
They're probably killing her. ELI She's right. He needs her alive. Xena enters. XENA What's going on? XENA Nothing. I'm fine. ELI We need to get to Kalib's as soon as possible. They move off. ELI Something's wrong. Kalib's always here. ELI No he likes doing things for himself. He's a real ascetic. We always called him the purist monk in his order. XENA Oh no. ELI About Kalib? What did he do with him? ELI Xena, is Kalib dead?
Xena pauses. XENA I believe so. ELI No it's not a waste. I won't let his death be a waste. There's gotta be something in these scrolls that explain all this! I'm gonna find it. XENA We'll help. Besides we've been on the run for two days. Go get some rest. I'll yell if I find something. Eli sits and Xena and Gabrielle leave. Joxer looks at the bath, but Xena is not there. He turns his back. Xena can be seen taking off her dress. She moves away.
Joxer enters where the bath is again. He turns his back to it and Xena gets in. Joxer turns around and sees Xena. He quickly turns around again and draws the curtain behind him. He stands with his back to Xena. I'm really sorry. I'll come back later. XENA No, no, it's all right. Has Eli found something?
Um, Xena you know how I feel about Gabrielle? XENA Sure, you love her. Um and um, I was kinda hoping that since you guys are such good friends and all that, you could maybe give me some tips. XENA Tips? I was thinking, uh, maybe I could try the down on one knee approach or uh, maybe flowers and--and uh, gifts or I could just club her over the head.
Do you see my problem? XENA Yeah, your expectations. XENA See right now it's all about getting a response from her. If ya love her, why don't you just tell her? Don't attach any strings to it. Make it unconditional. I tell her that and she says, "Joxer, what are you thinking? All right? Got it. Joxer leaves. XENA Of course, you could just kiss the girl. Xena chuckles. Ares enters. ARES Interesting offer. XENA I know you. You're the one who freed me from Kal. Ares stands behind Xena, rubbing her shoulders.
ARES Relax. Yeah, that's good. Just let go. It's amazing the difference when you don't resist me. XENA Amazing. ARES Open, free from old wounds, old grudges. XENA Very free. ARES And old ideas of right and wrong, good and bad, they just don't get in the way. XENA Bad? ARES Oops. I forgot. Neither does 'wrong' or 'evil. Ares stares at her chest. XENA You know what's strange? I know those words. I know their meaning , but somehow, they have no connection- She touches her chest.
XENA -here. ARES That's gotta be tough. He looks her in the eye. ARES But trust me, it's for a reason. But he doesn't know what it is.
Do you? Ares looks at her chest. ARES Oh, yeah. He looks her in the eye again. Not Kal, not even Zeus. Just the two of us, bringing peace and order to the world through force, something you and I were destined to do together. Ares moves to kiss her. Ares stops and Gabrielle enters. XENA Oh you know each other? Only you would take advantage of Xena when she's like this. ARES Like this? What do you mean? Because she can't remember things that make her mad? Things that fill her with hate?
Well, when did you become so hard? Gabrielle shakes her head. Xena looks upset and confused. What's Ares doing here? First Kal, then Ares.
So does Kal. What is it? ELI It's the same thing we need to get Xena back completely. Chakram of light. Now according to this only the purist soul can touch it without being destroyed. It's also one of the few things on Earth that can kill a god.
Whoever gets the chakram can wipe out all the other gods and make himself supreme. XENA Look, why don't we just leave it alone? Walk away? ELI Those gods would just keep trying until they got it. They all look at Xena. XENA All right.
Let's go for it. KAL So they think they've got it all worked out. How do you know? ARES You'd be amazed at what a little undercover work'll getcha. KAL Indulging yourself with a mortal like Xena is hardly work. My guess is your attempts to use her failed as miserably as my own. Ares holds his sword on Kal. KAL Cut the dramatics Ares. If Xena gets hold of that chakram, it could mean our lives. I say we call a truce 'till we stop her. Ares sheathes his sword.
ARES Okay, no more tricks. KAL So they're going for the chakram. Well let 'em try. We'll give them a reception they won't forget. She holds up the pieces of the chakram. XENA I can make no sense of this. She looks at the sky. I don't know how I'm supposed to be but the others say that I'm a warrior- Gabrielle enters from behind Xena. She remains silent and listens.
XENA -that I kill so that innocent people may live. They want me back the way that I was. Is that your plan? Send me a sign. How will I know what's right unless you do? XENA I can't help wondering if we're doing the right thing. Restoring this darkness, my violence, can that really be good?
I know that sounds strange coming from me. Over Why do we need to write down stories? What did people do before there were books and printing presses? There were still stories — they were told orally and passed on and people had better memories then. But we still have oral storytellers — I do oral storytelling, for example.
But as good as our memories used to be, they weren't good enough to remember all the stories. So isn't it wonderful that we have a way of keeping them. Otherwise, they'd all be gone forever.
How has the Internet affected your career as a writer? In some ways it makes me more accessible to my fans — this is good news and bad news. I do not yet trust the Internet for real research because in many instances I have no way of finding out if the research there is reasonable or accurate or true.
But with my own Web site, my fans or people who want to research about me will get the straight scoop. Do you think the Internet will affect the way we tell and pass along stories?
But again, good news bad news. Some is good in that we can connect with people all around the world. But some of it, and we have to learn to control it properly, is like Napster. Where they were stealing things that were written by people. I have found some of my stories and poems on the Internet without my permission, without payment, without my name attached to it — if that keeps happening, then people like me will have to stop writing. Because I make my living as a storyteller — and if people steal things from me, I can't afford to do it anymore.
Do you think today's child is forced to grow up too soon? I have grandchildren. I see certain things on TV or in books or in movies that I think are pushing ideas and concepts too soon onto them. I have to trust that their home life and their good common sense will win out in the end. Do you have a favorite passage from one of your books? It's dangerous to fall in love with one's own words. Do you come up with the titles before or after you write your books? Both ways.
Sometimes the editor suggests a title — sometimes we argue about a title until the last minute. And sometimes a title suggests a book. Are you working on any new books now?
I'm always working on new books. Right now I'm deep into a historical novel. I'm working on a book of poems with my son, who is a photographer. My daughter and I are working on our fourth book of our unsolved mysteries from history series, which will be about the Salem Witch Trials.
And I'm waiting for a revision letter for a novel called The Bagpiper's Ghost, and I'll start revising that when my editor sends me his comments.
And a music book with my other son, who is a musician. That's what I'm actively working on right now. And of course, the journal. Do you ever think about becoming something else other than a writer? Not anymore. I'm 61! But when I was younger, I wanted to be a ballet dancer, at another point I wanted to own a horse ranch, or be a lawyer. I have been an editor, a journalist, and a college professor. But I'm very happy now with my four jobs — I'm a writer, I'm a wife, I'm a mother, and I'm a grandmother.
What did you study in college? What is your favorite book that you have not written? That question can mean two things — the favorite book of mine that I have not written yet. One is a Guinivere book for young adults, and one is an adult book about the pre-Raphaelite brotherhood, which was a group of painters in England in the 19th century. My favorite book that I wish I had written — there are about a dozen of those. Is there anything else you would like to say to your readers?
Writing is like any athletics or dancing — it has to be practiced every day because otherwise the writing muscle goes flabby. But unlike dancing or athletics, you can be a writer, young middle-aged, or old. The more you do, the better you get. A writer doesn't run out of ideas — a writer runs out of time. Create a List. List Name Save. Rename this List. Rename this list. List Name Delete from selected List. Save to. Save to:. Save Create a List.
Create a list. Save Back. Jane Yolen Interview Transcript. Grades PreK—K , 1—2 , 3—5. Hence, we believe that the unexplored world of nORF peptides represents an untapped opportunity for discovery of new fundamental and translational areas of research. We hope this work will guide and motivate future detailed characterization of novel peptides in cancer and other diseases. Actively translated ORFs, having a footprints with clear sub-codon phasing or triplet periodicity, in each study included in this database, is detected systematically using the RibORF tool 53 , Further each ORF entry is annotated with its genomic position, strand, annotated ORF category canonical, truncated, extended, uORF, overlapping uORF, internal, external, polycistronic, readthrough, non-coding transcripts , length of encoded amino acid, ribosome profiling abundance RPKMs, raw read counts and the transcript to which the ORF maps probable transcript from which ORF is translated.
The samples were divided into 11 groups based on cell types. Every quantile had the same number of ORFs. The GENCODE v23 and corresponding Ensembl v81 genome annotations were downloaded, and transcript and coding sequence properties were extracted from the annotation files using a custom script.
All data processing was performed using R, R Studio, the R package Tidyverse, and unix command line tools. The Ensembl genome annotation was processed in R using ensembl db 56 , and genomic coordinates were processed using GenomicRanges. Set diagrams were produced using UpSetR. We included solid tumor TCGA cancer tissues with at least 50 samples, with matched NAT or GTEx normal tissue with at least 10 or 50 samples, respectively—a less stringent threshold for inclusion was used for NAT because these samples were less abundant.
RSEM expected count data was filtered to retain only selected samples and expressed transcripts prior to normalization and DE analysis. A single sample containing missing expected count values was excluded from this analysis. Prior to library size normalization and DE analysis, transcripts with poor expression were excluded from analysis. Applying a CPM threshold to identify expressed transcripts prior to TMM normalization and DE analysis has been shown to improve false discovery rate 57 and is recommended practice for edgeR.
Expressed transcripts are retained. Best practices for setting thresholds for transcript-level expression are poorly established, and the thresholds used in this study were, whilst informed by the literature, largely arbitrary.
To ensure consistent and reliable results, we included solid tumor TCGA cancer tissues with at least 50 samples, with matched NAT or GTEx normal tissue containing at least 10 or 50 samples, respectively—a less stringent threshold for inclusion was used for NAT because these samples are less abundant.
NAT samples closely resemble cancer samples both as a result of reduced variation in patient differences and sample processing. However, NAT is affected by changes in the tumor microenvironment and samples are less abundant than GTEx normal tissue samples. Seven cancer tissues included in this study are matched to both NAT and GTEx normal tissue which allowed us to determine whether DE results are reproducible across different reference tissues.
GffCompare was used to identify open-reading frames and transcripts with completely matching intron chains. GffCompare performs stringent filtering to detect and remove redundant input transcripts, and this deduplication is described in detail in the documentation. Novel and canonical ORF lengths were determined using ensembldb.
RNA-Seq expected counts were normalized across samples using the TMM 38 method to normalize for read depth and composition. As comparisons in DE were not made across transcripts, no normalization was introduced for effective transcript length.
To identify frequently expressed transcripts, CPM values were calculated across all expressed transcripts following TMM normalization using edgeR. Transcript DE was performed using all expressed transcripts to provide correct significance testing and improve reliability of dispersion estimation. The R package edgeR 39 was used to perform DE analysis using a GLM framework—this package was chosen as it is i highly cited, ii suitable for transcript-level analysis, iii compatible with non-integer expected counts from RSEM, and iv exhibits fast performance on large datasets.
A simple additive model with no intercept was constructed, with normal reference tissues and cancer tissues each represented by a single coefficient. No covariates, such as ethnicity, sex, age, or tumor grade, were controlled for in this DE analysis, but the GLM framework in edgeR was chosen because it would allow for control of covariates in follow-up analysis.
The process used for DE analysis is detailed in the edgeR manual. Briefly, transcript-wise dispersions were estimated under the GLM framework using the Cox—Reid profile-adjusted likelihood method, which takes into account multiple factors by fitting the described model.
A negative binomial model was fitted for each transcript, and thresholded hypotheses were tested to provide meaningful p -values and reliable control of false discovery rate. A fold-change threshold of 1. Coefficients representing cancer tissues and their corresponding normal reference tissues were compared under this framework.
The Benjamini and Hochberg method was used to adjust p -values for multiple testing and control false discovery rate. OS analysis was performed using the R packages survival 58 and survminer For each cancer type and for the nORF transcript considered, the cohort was split into high and low expression groups.
Kaplan—Meier curves were generated and curves were compared using a log-rank test. A Cox proportional hazards regression model was fitted to OS data and hazard ratios were derived from the model coefficients.
Both the Kaplan—Meier and Cox proportional hazards regression models assume proportional hazards, where the hazard ratio between the high and low expression groups remains constant over time. Protein domains were predicted from amino acid sequence using InterProScan We first curated a list of all nORFs that have been identified with evidence of translation. While all the other datasets contained protein sequences whose translation has been experimentally verified in literature, the downloaded RNAcentral dataset contained 9,, nucleotide transcript sequences.
We identified potential ORFs from these transcripts, using the following workflow. After removing redundant sequences from the extracted list, we obtained a unique set of 5,, protein sequences, which we used as putative transcripts from the RNAcentral database for disorder prediction. Since the size of the RNACentral dataset far exceeded that of the four other novel datasets, we decided to keep the datasets segregated for future analysis.
All statistical tests were corrected for multiple hypothesis testing, using FDR values computed by the Benjamini—Hochberg method. For each sequence, we predicted amino acid sites for nine PTMs—phosphorylation, acetylation, methylation, sulfation, SUMOylation, ubiquitination, C-linked, O-linked, and N-linked glycosylation. To test if each of the datasets NextProt, sORF, altORF, pseudogenes have higher or lower predicted PTM site densities than expected at random, we generated an individual control dataset specific to that dataset as follows.
We first obtained the average amino acid composition and length distribution for each dataset. We then fit a lognormal distribution to the sequence lengths.
Individual control AA sequences were then generated with lengths drawn from the lognormal distribution, and probability of each amino acid chosen from the average amino acid compositions for the dataset. We generated such control sequences until the control dataset had twice the number of sequences as the original dataset. ModPred was then used to predict PTM sites in these control datasets for the same list of nine modifications.
The number of predicted PTM sites in all datasets test or control were normalized to account for variable sequence length per residues. We investigated whether the pathogenicity scores of these mutations, assessed as combined annotation-dependent depletion CADD 33 and functional analysis through hidden Markov models FATHMM 65 scores, had any correlation with disorder scores at the mutated region of the novel proteins both amino-acid-specific disorder score, and average disorder score for a 7-aa window around the mutated residue.
This analysis Fig. Mice spleen tissues were obtained through a collaboration with Prof. All steps were carried out fast, and the cells were maintained in ice- and ice-cold buffers. We obtained on average 1. Cells bound with biotin-antibody, 1. Pellets were resuspended in 1.
Biotin-bound cells were depleted by passing through LD columns Milteny, in the magnetic field and flow-through was collected. A sample was prepared with all above five antibodies. Cells were processed through flow cytometer and following outputs were measured. CD25 channel filter removed 0. Gel lanes were cut into three sections for peptide extraction. After this period, the column valve was switched to allow elution of peptides from the pre-column onto the analytical column.
A genome index file to assist with read alignment was created using HISAT2-build, which extracts the exon and splice-site coordinates from the reference annotation. The StringTie merge function was used to create a list of non-redundant transcripts in B and T cells using the 12 sample-specific GTF files.
The master transcriptomic file was further analyzed as below. This filtering gave us , transcripts. The remaining transcripts were categorized into four sub-groups: B-male, B-female, T-male, or T-female, based on whether at least one out of three samples corresponding to a sub-group had a non-zero FPKM value.
Finally, the transcripts were categorized into B or T cell-specific transcriptomic datasets based on whether a transcript was present in at least one of the two sub-groups corresponding to a particular cell type.
This resulted in , B cell-specific transcriptomic dataset and 99, T cell-specific transcriptomic dataset. Transcript coordinates in the B and T cell-specific transcriptomic datasets were used to extract the corresponding nucleotide sequence from the reference genome using Bedtools Getfasta available in CGC.
SmProt contains a list of computationally predicted small peptides identified in several species including mouse. A macros code was, therefore, run on the SmProt website to specifically extract chromosome information for sORFs. Both databases had several duplicate entries which were removed by filtering them based on their chromosome location and amino acid sequence.
There are still a few sORFs in our database with the same chromosome coordinates, but these duplicates were not removed because their corresponding amino acid sequences were different. Few altORFs had multiple chromosome numbers assigned to it. Thermo mass spectrometry raw files were submitted to four databases search as described in Supplementary Fig. Briefly, an average of , mass spectra were obtained from each sample.
All mass spectra were initially searched independently against three amino acid databases—Uniprot database, sORF database, and altORF database and against the cRAP database of common contaminants. The enzyme specificity was set to trypsin, and two missed cleavages were tolerated. Carbamidomethylation of cysteine was set as a fixed modification, whilst variable modifications consisted of: oxidation of methionine, phosphorylation of serine, threonine, and tyrosine, and deamidation of asparagine and glutamine.
A minimum of two high confidence peptides per protein was required for identification. Out of , mass spectra, , mass spectra was mapped to Uniprot database; out of , mass spectra, 67, mass spectra was mapped to sORF database; out of , mass spectra, 32, mass spectra was mapped to altORF database. Finally, entries with no abundance values for all the four sub-groups were removed. All unmatched mass spectra from each step were then exported, combined into a single mgf and duplicates were removed.
B-cell-specific mgf file contained , spectra and T-cell-specific mgf contained , spectra. These files were then re-searched against B or T-cell-specific nucleotide proteogenomic databases in six frames. Spectral matches were then filtered and validated by two independent approaches. The first validation was done with Mascot Decoy analysis in Mascot and a second independent validation was done with Percolator analysis in Proteome Discoverer Thermo Scientific.
Transcripts that were only identified by both the validation methodologies and with at least two peptides matching them were considered as translated. A total of transcripts from both B and T cells nucleotide proteogenomic databases were identified to be translated with evidence of at least two peptides out of a total of peptides mapping to them and these regions were further analyzed as discussed below.
Of the transcripts identified to be translated, transcripts were identified in B cells and 86 transcripts were identified in T cells. These transcript regions varied in length, with the largest being 1.
So, we decided to investigate undefined ORFs based on individual peptides within these transcripts. To do this, we aligned the peptide and searched the genome up and downstream of the peptide until a stop or start codon was encountered.
Of these peptides we could annotate peptides into undefined novel ORF regions. A small portion of these ORFs could not be classified due to the genomic features from Ensembl disagreeing at different levels. Finally, the list of DE transcripts was filtered using Benjamin—Hochberg corrected p -values at a cutoff of 0.
This included the installation of the following software Hmmer suite 3. Using interproscan In order to allow a fair comparison, known proteins downloaded from uniprot with translational evidence were also annotated with interproscan serving as a reference point.
Analysis was performed on presence or absence of GO term annotation rather than the number of times the gene or protein might have been annotated with the same GO term.
Chi-squared tests were then performed with expected values based on the known protein proportions. The Bioconductor qvalue package was used to calculate q -values to be used for FDR correction.
For that we had to first identify human homologous sequences. LiftOver and tblastn mapped mouse sORFs to and regions for build hg38, respectively. Only regions for hg38 were mapped commonly from both LiftOver and tblastn.
Mapped mutations from each region were then compared to the coding sequence of each sORF to determine potential changes to amino acid sequence using python script. For the sORFs with predicted structures available, the mutations were mapped onto the PDB file and visualized with Pymol as red colored residues.
The expression of transcripts translating low-noise nORFs identified from datasets corresponding to 11 cell types downloaded from RPFdb , was investigated in 19 human cancers with the objective to identify probable cancer markers.
The TCGA cancers and their corresponding healthy tissues from GTEx, along with the number of samples in each case, analyzed in this study is given in Supplementary Table 6.
Human ortholog transcript of one mouse sORF that is translated in mouse B and T cells was identified, its structure predicted and inhibitors were screened against it. The details are as follows. We mapped two cosmic noncoding mutations to this transcript. Figure 6b shows the structure along with the mutations mapped. Structure predicted from ENST First in the protein preparation step, the structure was minimized using protein preparation wizard in maestro The grid was generated at all the active site residues of the topmost scoring pocket identified by the two tools.
Peptide sequence of the product translated from ENST The predicted active Site Residues used in docking are given in Supplementary Table 6 and the accompanying associated figure is Supplementary Fig. The virtual screening involved the following three stages: 1. Listed below are the details of the predicted best hit compounds searched from the three asinex libraries. Docking scores for the top Immuno-oncology library compounds, targeted-oncology library compounds, signaling pathway inhibitors are given in Supplementary Tables 7 — 9 , respectively, and their associated figures are Supplementary Figs.
MM-GBSA-binding energies, which estimates relative binding affinities for the few best hit immuno-oncology compounds, targeted-oncology compounds, and signaling pathway inhibitors are given in Supplementary Tables 10 — 12 , respectively. Almost all processed data is in the main text or in the supplementary materials. Vitting-Seerup, K. The landscape of isoform switches in human cancers. Cancer Res. Hu, X. The role of long noncoding RNAs in cancer: the dark matter matters.
Rheinbay, E. Analyses of non-coding somatic drivers in 2, cancer whole genomes. Nature , — Wang, J. Brunet, M. OpenProt: a more comprehensive guide to explore eukaryotic coding potential and proteomes. Nucleic Acids Res. Plaza, S. In search of lost small peptides. Cell Dev. Clamp, M. Distinguishing protein-coding and noncoding genes in the human genome.
Natl Acad. USA , — Prabakaran, S. Quantitative profiling of peptides from RNAs classified as noncoding. Ruiz-Orera, J. Translation of neutrally evolving peptides provides a basis for de novo gene evolution. Zhu, Y. Discovery of coding regions in the human genome by integrated proteogenomics analysis workflow. Olexiouk, V. Using the sORFs. Org Database.
0コメント