Details
-
Type:
Task
-
Status: Closed (View Workflow)
-
Priority:
Major
-
Resolution: Done
-
Affects Version/s: None
-
Fix Version/s: None
-
Labels:None
-
Story Points:2
-
Epic Link:
-
Sprint:Spring 9 2022 May 9, Summer 1 2022 May 23, Summer 2 2022 June 6, Summer 3 2022 June 21, Summer 4 2022 July 4
Description
The S. lycopersicon (cultivated tomato) gene annotations include only one gene model per gene. However, visualizing RNA-Seq data in IGB shows that a large number of genes produce multiple splice forms. At least one other group has noticed this, as well. In their article "Expanding Alternative Splicing Identification by Integrating Multiple Sources of Transcription Data in Tomato", a group at Ohio State University led by Prof. Xiangjia (Jack) Min reported using transcriptome data, including ESTs and RNA-Seq data, to assemble new gene models. I downloaded these and deployed them to IGB Quickload; they are one of the available data sets for the next to last genome release.
There may be other groups developing similar datasets for the most recent genome release for tomato. And in order to quantify splice variant expression using current methods, it would be extremely helpful to have an up-to-date, accurate-as-possible collection of gene models annotated with functional information. Who else is interested in this and would be interested in contributing? Or, is this something only our group might care about?
As part of the pollen NSF project, we are trying to understand and discover how heat stress triggers changes in RNA synthesis in pollen, in pollen tubes, and in other sample types related to reproduction in plants, especially tomato?
How homogenous are the RNA-Seq data sets coming from the pollen project? So far, all the data have been from a single cell type: germinating pollen tubes. I do not recall seeing much evidence for alternative splicing in these datasets, at least not as compared with other samples that included many cell types, e.g., root or shoot. Also, are there splice forms that exist mainly in pollen but not other tissue types? We found some examples of this in the Arabidospis pollen RNA-Seq data described in our paper "RNA-seq of Arabidopsis pollen uncovers novel transcription and alternative splicing".
How many tomato RNA-Seq data sets are there, and how good are they? For the purpose of producing new gene models, the best bulk RNA-Seq data would be paired end, very long read lengths, and strand-specific. Are such data available currently, or would we need to create new data to cover the entirety of transcription?
Attachments
Issue Links
- relates to
-
-
- Closed
-
Activity
A paper was published in 2019 that discussed using RNA-Seq data harvested from the public domain (e.g., SRA) to form a new set of gene models for tomato, using the SL3.0 (feb 2017) assembly for generating alignments, which were used to form gene models. [~aloraine] set up those gene models for visualization in IGB as described in the README.md here: https://bitbucket.org/hotpollen/rna-seq/src/master/SplicingBackground/.
Update:
- We looked at the spreadsheet during a meeting (11:00 am Tues 6/22).
- Labeled (using color) the experimental (target) dataset (green); candidate positives controls (grayish-blue); picked first positive control dataset (SRP328042)
- Decided to filter candidates on +1 or -1 the target datasets base length (filtered on paired-end already)
- Made folders in Google Drive to track today's products
- Downloaded automatically-generated file containing 18 "SRR" run accessions in SRP328042.
- Decided next steps:
Next steps:
- Download fastq format files from sequence read archive, put onto cluster (using fasterqDump from SRA toolkit)
- Retrieve data for experimental () and positive control (SRP328042)
- Make sure that the Unix file permissions are set such that everybody in the group can edit, change names, delete the files
- Please provide a simple script (nothing fancy) that will do the above
Probably next steps:
- Develop configuration files for running rna-seq data analysis pipeline from nextflow (nfcore / rnaseq)
Molly: Video on working with GEO (not SRA) in R, bioconductor, tidyverse is here: https://youtu.be/dc77edcNp3M
Today we learned about a new tomato genome release with what appear to be splice variants.
Loaded GFF file into IGB (w/o the sequence) and clicked on a few things.
Questions about the gene models :
- How well represented are genes expressed primarily or exclusively in germinating pollen, as compared to other, more exhaustively sequenced organ or tissue types?
- How are transcripts assigned to the same parent identifier?
To-Do for splicing investigation project, now that we have these splice variants (maybe?)
Start testing splicing detection methodologies, starting with ArabiTag, as this is code Ann is very familiar with.
To run ArabiTag:
- Convert the gene models into bed12 format, use ArabiTag to detect and classify the AS event pairs (will probably use: https://bitbucket.org/lorainelab/genomesource/src/master/)
- Run ArabiTag on read1 and then read2 of experimental and positive control datasets to generate counts for S and L forms of each pair (arabitag: https://bitbucket.org/lorainelab/altspliceanalysis/src/master/)
- Analyze read1 and read2 results separately, not together, to avoid fragments double-sampling when R1 and R2 overlap
- Run ArabiTag R code (via Markdown) to calculate %spliced-in, t-statistic for comparing two groups, counts per comparison, false discovery rate (Q) per comparison, and more (for examples, see: https://bitbucket.org/lorainelab/hot-dry-arabidopsis/src/master/ and https://bitbucket.org/lorainelab/ricealtsplice/src/master/)
- Select genes with RNA processing or RNA-binding annotations
- Visually analyze the data by viewing RNA-Seq sequence alignments in an interactive genome browser
- Compare the amount of splicing difference in the positive control versus the experimental sample. Are they similar? Are they different?
Methodology question:
Can we merge overlapping R1 and R2 sequences into a single sequence and align the entire thing, rather than the two ends separately? Somehow, that seems better.
Assuming one has downloaded the SRA toolkit locally.
SRA fasterqdump:
~/sw/sratoolkit.3.0.0-mac64/bin/prefetch SRR1572591
fasterq-dump -S SRR1572591.sra
(The -S makes the split into 2 fastq files)
Now validate what you pulled:
~/sw/sratoolkit.3.0.0-mac64/bin/vdb-validate SRR5790104
Saw talk at meeting which mentioned this new method for transcript assembly: https://genomebiology.biomedcentral.com/articles/10.1186/s13059-022-02700-3
Slides from meeting review:
https://docs.google.com/presentation/d/1YmZWDT8-COLkEtj3ppA7HadD_Zyy6uD_I3V6K90jmjg/edit#slide=id.p
More questions:
What are the splice variants of tomato?
What are the most frequently observed transcript models for each gene model?
What conditions change the balance? Which coditions or treatments push the observed transcript ratios out of balance?
What is "balance", and how would it manifest in the biological data we've collected?
Possible strategies for answering the above questions:
Possible ways to do the above:
Then, describe as plainly as possible what was found.