Available from Version: 1.0
miRNAs are a well-studied class of sRNAs that are generated from a single-stranded RNA (ssRNA) that forms a stable, partially double stranded stem-loop structure (hairpin). miRCat predicts miRNAs from high- throughput sRNA sequencing data without requiring a putative precursor sequence as these will be identified by the program.
miRCat projects are created by entering the new project menu which is found under the file menu.
Once the sequences are mapped to the input genome, miRCat will look for genomic regions covered with sRNAs (sRNA loci), containing reads with abundance at least five (this threshold can be adjusted using the min abundance parameter).
These loci must match certain criteria:
- Loci must contain no more than four non-overlapping sRNAs.
- Each sRNA in a locus must be no more than 200nt away from its closest neighbour (this threshold can be adjusted using the –hit dist parameter).
- At least 90% of sRNAs in a locus must have the same orientation (this threshold can be adjusted using the –percent orientation parameter).
Once a list of loci has been produced, it is further analysed in order to find likely miRNA candidates:
- The most abundant sRNA read within a locus is chosen as the likely miRNA.
- Flanking sequences surrounding this sRNA are extracted from the genome using varying window lengths.
- Each sequence window is then folded using RNAfold, producing a secondary structure for the putative miRNA that can then be viewed using the Hairpin Annotation Tool
- miRCat then trims the secondary structure and computes discriminative features useful for classifying miRNAs. The features are:
- The number of consecutive mismatches between miRNA and miRNA* must be no more than 3.
- The number of paired nucleotides between the miRNA and the miRNA* must be at least 17 of the 25 nucleotides centred around the miRNA.
- The hairpin must be at least 75nt (for plants) or 50nt (for animals) in length.
- The percentage of paired bases in the hairpin must be at least 50% of base-pairs in the hairpin (this threshold can be adjusted using the –max percent unpaired parameter).
- The hairpin with the lowest adjusted minimum free energy (AMFE) from the sequence windows is then chosen as the precursor miRNA (pre-miRNA) candidate
- The pre-miRNA candidate is then tested using randfold using a AMFE cutoff.
Hairpin Annotation Tool output showing miR164 precursor.
Parameters:
- genomehits: The maximum number of genome hits. (1 > genomehits, default genomehits = 16).
- hit dist: The maximum distance between consecutive hits on the genome. (hit dist, default hit dist = 200).
- max gaps: The maximum number of consecutive unpaired bases in miRNA region. (0 > max gaps < 5, default max gaps = 3).
- max overlap percentage: The maximum total percentage of miRNAs that lie under the miRNA or miRNA* position on the hairpin. (30 < max overlap length, default = 70).
- max percent unpaired: The maximum percentage of unpaired bases in hairpin. (1 < max percent unpaired < 100, default max percent unpaired = 50).
- max unique hits: The Maximum number of non-overlapping hits in a locus. (1 < max unique hits, default max unique hits = 3).
- maxsize: The maximum length of a miRNA. (18 < maxsize < 24, default maxsize = 22).
- min abundance: The minimum sRNA abundance. (1 < min abundance, default min abundance = 5).
- min energy: The adjusted minimum free energy of the hairpin. Must be < 0. Default = -25.
- min gc: The Minimum percentage of G/C in miRNA (must be > 1 and < 100. Default = 10).
- min hairpin length: The minimum length of hairpin (nt) (must be > 50. Default = 75).
- min paired: The Minimum number of paired bases in miRNA region (Must be > 10 and < 25. Default = 17).
- minsize: The Minimum sRNA size (Must be > 18 and < 36. Default = 20).
- Allowing complex loops: This will allow or remove any hairpins containing complex loops.
- Orientation percentage: The percentage of sRNAs in locus that must be in the same orientation (1 < percent orientation < 100, default percent orientation = 90).
- Hairpin Extension: How much each hairpin should extend past the sRNA read to form the window (40 < window length < 400, default window length = 150).
During the analysis procedure the results are entered into the table as shown below:
The table contains the following information:
- Chromosome
- Start position
- End position
- Strand/orientation
- Abundance (number of times sequenced in high-throughput dataset)
- Sequence of predicted mature miRNA
- Representative sequence accession from input dataset
- Length of predicted mature miRNA
- Number of matches to genome
- Length of predicted precursor hairpin sequence
- G/C % content of hairpin sequence
- Minimum free energy (MFE) of predicted hairpin sequence
- Adjusted MFE, AMFE = (MFE / length of hairpin) * 100
- Shows MFE per 100nt making results comparable
- miRNA* shows predicted miRNA* sequence(s), if any, along with abundance in input dataset shown in brackets
- Hairpin Sequence (with miRNA sequence highlighted in blue and miRNA* if present in red)
- Hairpin Dot-Bracket notation
- Hairpin start position
- Hairpin end position
- miRNA* start position, if present
- miRNA* end position, if present
A user has the option to interact with the results in real time in several aspects. Using the Export menu a user can export the results to file. A user can also use the controls shown at the top to output the results as they are generated to file or pipe all results into into either the RNA Folding/Annotation tool or the VisSR tool.
The miRCat toolbar
Additionally a context menu has been included to allow the user to pipe a single result line into two other tools in the sRNA Workbench.
The miRCat context menu
Both options operate on the currently selected result line. ‘Render Hairpin’ will render the selection in the Hairpin Annotation tool, while ‘Show in VisSR’ will display the selection in VisSR.
The image below shows the VisSR output of a single miRNA classified in mircat:
VisSR showing miRNA 165 after being classified with miRCat
Hi Matt,
I have a quick question: does miRCat map the reads to the input genome searching only for perfect matches or allowing mismatches?
Thank you very much!
Best wishes
Alice
Hi Alice,
miRCat only allows for putative miRNA (or other small RNA) that have an exact match to the genome to be added to the locus. You could probably force something different to happen if you need to, just align the sequences using the sequence alignment tool and choose your mismatch options, then use the file it creates as input to miRCat rather than loading the sRNAs directly…
This is completely untested on my end though! You may end up with a huge amount of results this way!
I hope this helps
Best wishes,
Matt
Hi Matt,
I am interested to the reads that map to the genome with an exact match, I only wanted to be sure of that!
Thank you very much!
Alice
Hi Matt,
I also tried to use miRCat, the process seems to run but when it is complete I don’t get the results, I see “Status: completed successfully” but the screen is empty, do you have suggestions for that problem?
Thank you!
Alice
If you want I can write you a private e-mail in which I give you the username and the password of my account in the server, maybe if you make a try with my data it would be easier to understand what is the problem (I tried in another server the same tool and it gave this error message: “input string: 2,78″ or “input string: 3″)
Thank you very much for your help!!
Alice
Comment resolved now privately, topic is closed
Hi,
I’m using v2.4 workbench – I have exported the miRCat output to a .csv file. I have two questions:
1. The input sRNAs were already made non-redundant. I am seeing a discrepancy between the ‘Genomic Hits’ field, and the actual number of miRNAs listed/detected in the .csv file. For example, a particular sequence is reported to have 13 ‘Genomic Hits’, but I can only find 5 genomic loci in the same file. I’m seeing this for many sequences. What is the cause for this?
2. How does miRCat deal with sequences that are a ‘subset’ of a longer sequence? e.g. a 21 nt miRNA that has the same sequence as a 22nt miRNA? Are these classified as separate?
Many thanks,
Ken
I just read a comment below about the number in the brackets in the fasta file output being genomic coordinates, which is what I am seeing in my fasta output too. Perhaps my question 1 will be solved in the latest version. I’ll try the latest version to confirm this.
Hi,
Yes this bug was reported and fixed in version 2.5.0. Please refer to the notes below:
http://srna-workbench.cmp.uea.ac.uk/the-uea-small-rna-workbench-version-2-5-0/
cheers,
Matt
Hi,
1: Just need a little clarification on which bits of info you are looking at, Genomic Hits are the number of times the individual sequence aligned to the genome (i.e. if you used the sequence alignment tool on its own, you would find that sequence n number of times in different locations) is the value you are referring to in the CSV the number of times that particular read was sequenced? Or the number of times it was reported as being a miRNA in the results?
2: Do you mean specific isomirs that have been sequenced here? we currently treat each sequence as having the potential to be a miRNA and do not look to see if they are isomirs of another miRNA (but it is an interesting thought! perhaps I can look to include something like this! It is considered in miRProf already)
cheers,
Matt
Hi Matt,
1. The value I am referring to is found in the ‘Genomic hits’ column in the exported CSV file, which is column 14 (by the way, the columns in the actual CSV file do not correspond to the explanations given on this page above). I am assuming this is the number of times that the particular predicted miRNA aligned to the genome. It is not the read abundance. Actually, I may have just answered my question … I assumed that the number of genomic hits would equate to the number of genomic loci reported for the particular miRNA, but this is not the case I think? So if the particular miRNA had 13 genomic hits, perhaps only 6 of these hits had evidence to actually support the sequence to be a true miRNA, which is what is reported in the CSV file? Put another way I found 6 instances of a miRNA sequence in the CSV file, yet the number of genomic hits was reported to be 13.
2. Yes I am referring to isomirs. It would be good to know the %’s of these within the dataset – I have around 3900 potential miRNAs from miRCat, but it would be good to know how many actual families there are. I guess I could use the fasta output from miRCat, make them redundant again based on the read counts, and use this as input to miRProf to just get the classifications. However, I would miss novel families. Do you have a tool to just detect isomirs?
cheers,
Ken
Hi,
1) Genomic hits will be the exact number of times that sequence aligned to the genome, however, as you have suggested, not all of those hits might be considered a miRNA if (for example) the flanking region around it did not fold into a hairpin (there could be other reasons it was not classified in that location as a miRNA) the program does not take into account if a particular sequence has already been classified.
2) miRProf should be able to read your non-redundant output from miRCat provided you are using the workbench version not the perl/online version. This should give you the information on which sequences belong to which family etc. There is not a tool that will tell you specifically which sequence are isomirs of each other but miRProf will group all sequences that align to a certain miRNA found in miRBase (you can see them all by dropping the arrows down in the interface) however, you will find the output CSV from miRProf is not quite functioning correctly at the moment (I am fixing it for the next release) but you can see them all in the GUI output
Cheers,
Matt
hii Matt,
i am using mircat tool in sRNAWorkbenchv2.3.1and i had done adaptor removal,filter and sequence alignment.i have used input as .patman file in mircat and select default parameter of plant. now,i am getting error in mircat like “mircat connection lost” then if i am trying next time then error like “this port is already in used”.
What am i supposed to do?
thank you.
Hi,
More than likely the program did not close correctly at some point and some zombies are hanging around on the ports the software wants to use. Could you kill any java processes related to your login? Depending on OS this can be done with a single command. Let me know if you need help with this!
After killing any java processes attempt to run the software again and let me know if it works for you
Cheers,
Matt
Hi,Matt,
I am not quit sure about the mean of the mirCat output,please give a hand.I had our smRNA data processed by mirCat and got the output,while when I export the miRNA to files,I got file like this:
>TCGGGCCAGAGATTCGGACCTT(74)
TCGGGCCAGAGATTCGGACCTT
>TTGACAGAAGATAGAGAGCAC(134)
TTGACAGAAGATAGAGAGCAC
>GATGAACATTAGCGCTAGAGCTGA(649)
GATGAACATTAGCGCTAGAGCTGA
what is the mean of the number in the title,and how do you get it,can I directly compare it among different smRNA libs?
Hi Yan,
The value in the brackets is the non-redundant count of that sequence. All FASTA files are converted to (if not already) a non-redundant format prior to processing, so the sequence
TCGGGCCAGAGATTCGGACCTT
appeared 74 times in your original input set and as such was selected as the most abundant small RNA within that locus (therefore used for further testing for miRNA viability, all of which it passed).
To directly compare sequence abundance among different libraries you will first need to normalise the abundance data.
I hope this helps!
Matt
I just ran some mirCat analyses and in the mature miRNA fasta files the number in brackets are actually the start coordinate in the genome and not the abundance (I checked this by comparing to the GFF and the results in the csv file). Is this a known bug?
Is certainly is! But one I spotted only a couple of days ago.. I usually just go by what is in the table until I output some FASTA for one of the wet lab guys downstairs and thought…. hmmm that abundance is larger than the entire sample!
I have fixed it and included it on my forthcoming release which is close to being ready
Thanks for letting me know though!
Best wishes,
Matt
Hi Matt,
I noticed a problem,I use the same data and same parameters to run mirCat ,while sometimes I could not get the same rusults ,for example,once I got 593 predicted miRNAs,and once I got 580 miRNAs.
I think this maybe caused by fail output of some thread,if so ,how should I solve this problem? Here are some information warning during the run:
Dec 10, 2012 8:10:05 PM uk.ac.uea.cmp.srnaworkbench.utils.WorkbenchLogger log
SEVERE: WORKBENCH: java.lang.String.substring(Unknown Source)
uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.getStructure(Window.java:1403)
uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.processSingleWindow(Window.java:741)
uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.processWindows(Window.java:318)
uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.run(Window.java:121)
java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source)
java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
java.lang.Thread.run(Unknown Source)
Dec 10, 2012 8:10:29 PM uk.ac.uea.cmp.srnaworkbench.utils.WorkbenchLogger log
SEVERE: WORKBENCH:
java.lang.StringIndexOutOfBoundsException: String index out of range: -1
at java.lang.String.substring(Unknown Source)
at uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.getStructure(Window.java:1403)
at uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.processSingleWindow(Window.java:741)
at uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.processWindows(Window.java:318)
at uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.run(Window.java:121)
at java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
at java.lang.Thread.run(Unknown Source)
Dec 10, 2012 8:10:29 PM uk.ac.uea.cmp.srnaworkbench.utils.WorkbenchLogger log
SEVERE: WORKBENCH: java.lang.String.substring(Unknown Source)
uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.getStructure(Window.java:1403)
uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.processSingleWindow(Window.java:741)
uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.processWindows(Window.java:318)
uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.run(Window.java:121)
java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source)
java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
java.lang.Thread.run(Unknown Source)
Dec 10, 2012 8:11:41 PM uk.ac.uea.cmp.srnaworkbench.utils.WorkbenchLogger log
SEVERE: WORKBENCH:
java.lang.StringIndexOutOfBoundsException: String index out of range: -1
at java.lang.String.substring(Unknown Source)
at uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.getStructure(Window.java:1403)
at uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.processSingleWindow(Window.java:741)
at uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.processWindows(Window.java:318)
at uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.run(Window.java:121)
at java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
at java.lang.Thread.run(Unknown Source)
Dec 10, 2012 8:11:41 PM uk.ac.uea.cmp.srnaworkbench.utils.WorkbenchLogger log
SEVERE: WORKBENCH: java.lang.String.substring(Unknown Source)
uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.getStructure(Window.java:1403)
uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.processSingleWindow(Window.java:741)
uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.processWindows(Window.java:318)
uk.ac.uea.cmp.srnaworkbench.tools.mircat.Window.run(Window.java:121)
java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source)
java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
java.lang.Thread.run(Unknown Source)
Hi Yan,
Yes as you correctly say, this is most likely due to a fail on one of the process threads. It is difficult to say exactly why this happened, it could be a program error/bug or a problem with the input data. You could try downloading the latest version of the program and reducing the thread count down to just 1 and see if the error persists (this will of course increase the run time) or send me a small sample of the data and I can look at it and see if I can re-create the problem
Let me know!
Thanks,
Matt
Hi Matt,
We’ve made a transcriptome dataset, and now I’m trying to annotate it for miRNAs. We’re using 454 reads from a non-normalised library which have already had their adaptors removed. My main problem is that when I try to run filter or mirCAT it either returns no valid sequences; or that my data file may be non-redundant. Is this an incompatibility with the program and 454 for example?
P.S. I’m using the web-based server.
Kind Regards,
Angus
Hi Angus,
I believe the web version does require the files in non-redundant format (the downloadable version can handle both). However, 454 reads should be ok. Could you send me the first few lines of your input files and I will see if I can work out what the problem might be?
matthew.stocks@uea.ac.uk
Thanks,
Matt
Hi,
Thanks Matt, I’ve sent you the first lines,
A
Hi Angus,
Ahh now I see what you mean, sorry about that!
Ok, miRCat requires small RNA reads to predict miRNA sequences and will not annotate transcript data. Did you sequence small RNA sequences in parallel with this experiment? If not, the best you can do is use BLAST to annotate the known miRNAs within the transcriptome.
Hope this helps,
Matt
Hi,
There are sRNA reads within the data set. When filtered with the Filter tool they do get flagged up (about 2000 in the 23nt range), but these cannot be mapped against any genomes or even the entire miRNA data set. The program suggests either adaptors have not been removed or that I am mapping to the wrong genome, but that can’t be a problem because it’s being mapped against everything.
Cheers,
A
Hi Angus,
Apologies for the delay in replying to this I have been away on annual leave last week.
I would say that although it is possible you have some small RNA reads in your data, you would be very lucky to find a real micro RNA with only 2000 sequences to search through. (also, I believe the web based version of miRCat will discard all reads with an abundance of 1 but this can be configured by the user in the downloadable workbench if you wish to bypass this behaviour)
My suggestion would be to run the small RNA reads you have through miRProf and this will at least tell you if you have any known micro RNA within your data. If you have none, then it is unlikely you have any new ones also.
In regards to the mapping issue, are you sure you have the correct adapter sequence? Did you receive any information on which adapter was used during sequencing?
Thanks,
Matt
Hai,
I was using miRCat for analyzing my small rna sequenced data. Where could I get a complete Arabidopsis genome sequence.Also, while removing adaptar sequences using helper tool, I am getting out of memory error. My system is having 8GB RAM. I would like to have your suggestion to overcome this problem.
Hi Kavitha,
the best location for the A.TH genome is the TAIR website:
ftp://ftp.arabidopsis.org/home/tair/Genes/TAIR10_genome_release/TAIR10_chromosome_files/
the full genome is the file called TAIR10_chr_all.fas (or you can download each chromosome separately from here also)
@Overcoming the memory issue:
It depends, how large is your initial input file? Is it a single A.th sample? How are you running the workbench? Did you start it using the sRNAWorkbenchStartup.jar? Which OS are you using? 8GB of RAM is usually enough for even the largest files, did you have a lot of other memory intensive programs running at the same time?
Dear Sir,
Thank you for the reply.My initial output file is 7.7Gb.Yes I am using a single sample. I am starting it using sRNAWorkbenchStartup.jar. I am using windows 7. While using the small rna workbench,I am not running any other programs.
Ok that could be too much data for that amount of memory. Could you send me the first 100 or so lines of the small RNA file that you input into the Adapter Removal tool? So I can check the data looks ok.
send to: matthew.stocks@uea.ac.uk
You may need to strip the adapters using the web based service if you do not have access to a server or PC with slightly higher memory capacity to run the workbench from
Hello,
I am using this workbench for analyzing my small RNA sequencing data. I have a question. From where, I can download the genome file. I am using rats for my experiment. I tried to download from ensemble, but it has many files corresponding for each chromosome. But it seems in miRcat, we need a single genome file. How to achieve this?
Thank you
Sincerely,
Velmurugan
Madurai Kamaraj University, India
Hi Velmurugan,
miRCat will work with single chromosome files or the entire genome, sometimes it is better to work with single chromosomes for certain data because the entire genome file is too large to fit within memory. For example, people working with human data often just map their small RNA reads to a single chromosome and then use that file for there analysis (or you can map your reads to the entire genome using our sequence alignment tool and then use single chromosome files within miRCat.
If you want to use an entire rat genome file you can find one here:
ftp://hgdownload.cse.ucsc.edu/goldenPath/rn5/bigZips/
there is a web page describing each of the files within the directory linked above
http://hgdownload.cse.ucsc.edu/goldenPath/rn5/bigZips/
and the file with all the chromosomes is called rn5.2bit or this is also a fasta gzipped download called (rn5.fa.gz) which will contain the same information.
Please feel free to let me know if you need any further help!
Best Wishes,
Matt
Hi
I have pre miRNA sequences of plants,can i have primary miRNA sequences from it by computational method.Also suggest suitable publication / literature.
Hi,
Please refer to my reply below to Chintan Vora regarding pri-miRNAs and detection within miRCat.
In addition, a possible common motif in pri-miRNA was detected and mentioned in Bartels talk on “Systems Approaches to miRNAs” at Keystone Symposia on Molecular and Cellular Biology: Gene Silencing by small RNAs this year:
http://www.keystonesymposia.org/meetings/viewMeetings.cfm?MeetingID=1177
Hope that helps.
Kind regards,
Matt
Can we change the adapter sequence in miRCat other than the default 3 sequences provided?
Yes you can change the adapter sequence after enabling the Adapter Removal panel using the check box. Simply type the desired 3′ or 5′ adapter sequence into the box(es) provided in the interface (just above the drop-down menu where you can select the provided sequences.) The drop-down menu is just there for convenience. Alternatively you can access all of these options from the stand alone Adapter Removal tool.
Can we identify polycistronic miRNA using sRNA workbench?
Yes miRCat can identify pre-miRNAs from the same primary transcript but does not attempt to classify pri-miRNAs (and will therefore treat each pre-miRNA as a separate entity). If you need any further information please dont hesitate to ask!
Thank you for the reply but i am interested in identifying polycistronic miRNA(pri-miRNA). Can you suggest any tool or publications which can list the features of pri-miRNA.
Usually pri-miRNAs are identified based on clustering pre-miRNAs based on distance between loci. E.g. miRBase uses an arbitrary threshold of 10Kb, any miRNAs within that distance are grouped into the same putative pri-miRNA. As far as I know there are no published computational methods that will accurately identify pri-miRNAs and no way of telling from small RNA sequencing what the start/end of these primary transcripts is.
miRCat may miss miRNAs that are in close proximity to one another as it will group all neighbouring sequences together into loci before folding them and analysing the secondary structure (therefore two different miRNAs may be joined into one locus and will fail the structure prediction step). For this reason you may need to lower the “locus separation distance” in order to differentiate between miRNAs within close proximity to one another.
How do i update mirbase in mircat?
Hi,
RE your problem with miRBase (copied from another response)
you may be experiencing an issue related to Java 1.7 when attempting to download/update miRBase files (related to support for IPv4, Java are aware of the problem but I am not sure when their updated code will be available). We are attempting to add a work-around at the moment.
The software should work fine on Java 1.6 (any update). If you want instruction on how to use an earlier version of Java please let me know.
Please let me know if I can be of any more help,
Thanks,
Matt