Tablet - Next Generation Sequence Assembly Visualization

[imilne]

Hi all,

Some of you may be aware of this already, but here at SCRI we've been working on producing a visualization tool for NGS assembly data called Tablet. Our main goal was to create something that's not only fast and provides nice visualizations, but also be both easy to use and easy to install.

Tablet was designed to aid our work here, which is primarily on plant data, but we've had some success loading in some human stuff too (just for fun obviously!).

We've attempted to get working parsers for a wide range of common formats (ace, afg, maq, soap) and we currently have experimental support for sam too (until we can get the Picard API working properly).

Please give it a try, and let us know what you think, either here, or by emailing us directly (there's an option within Tablet itself to do this). It's available in both 32 and 64bit versions, for all the usual suspects (Windows, OS X, Linux, Solaris), and can be downloaded from the following URL:

http://bioinf.scri.ac.uk/tablet

We also have a paper in advance (open) access with Bioinformatics, which is linked to from the above.

Tablet is still very much work in progress though, so do feel free to suggest any enhancements and improvements you'd like to see.

Iain

(ps: I think these attached pictures are going to be tiny, but the full versions are on the web site)

[Zigster]

this is very slick and unlike every other short-read viewer I have ever used it worked right out of the box

[lh3]

For human data, it would be essential to support BAM natively. By "natively", I mean to directly load BAM alignment without converting it to other formats. Most human alignments are available in BAM on NCBI ftp. It is awkward and inefficient to convert a compressed file of >100GB. In addition, tablet seems not supporting insertion in SAM/BAM probably because unlike ACE/CAF, the viewer must be able to compute padding by itself.

Except these, tablet looks great!

[maasha]

How does it compare to magicviewer:

http://bioinformatics.zj.cn/magicviewer/

?

Martin

[imilne]

For human data, it would be essential to support BAM natively. By "natively", I mean to directly load BAM alignment without converting it to other formats. Most human alignments are available in BAM on NCBI ftp. It is awkward and inefficient to convert a compressed file of >100GB. In addition, tablet seems not supporting insertion in SAM/BAM probably because unlike ACE/CAF, the viewer must be able to compute padding by itself.

Except these, tablet looks great!

Yeah, we're aware of the issues with BAM. We've been having trouble getting the Java API for SAM/BAM working properly (which is why we knocked out a quick and basic SAM-only parser). We're not sure yet what the problem is: Picard might be too strict on the files it accepts, some of the assemblers might not be producing valid files, we might not be using the API correctly, etc. To complicate matters further, we don't actually have anybody here at SCRI generating data in SAM/BAM format either.

Things have been looking a bit more hopeful this week though, so maybe we'll have it working soon...

[imilne]

We now have a new release out - 1.10.01.28 - with the following new, changed, or fixed features:

- NEW: Added a right-click "save a summary of read information (per contig)" option to the contigs table.
- NEW: Tablet can now read from http streams (supported on the command line and the Open Assembly dialog).
- NEW: Tablet can now read from compressed gzip data streams (http or file).
- NEW: Features outside the scope of the current contig are now highlighted in red in the Features Table.
- NEW: Features with a Name= tag in a GFF3 file will now show this name in the Features Table.
- NEW: Added an option to toggle on or off the use of regular expressions in searches.
- CHG: The Importing Assembly dialog now shows its progress in MB/s too.
- CHG: Removed support for consensus tags in ACE files. GFF3 formatted files are now the only way to import features.
- BUG: Loading the same features file twice will no longer result in duplicate entries.

Support for BAM is also progressing well, and should be available in the next release if all goes to plan.

Iain

[imilne]

Another update out today - 1.10.02.08

- NEW: Added (experimental) support for BAM, using in-memory only reading for now, NOT indexed support.
- NEW: Support for most CIGAR operations, with the positions of Insertions being tagged as features for now.
- NEW: Displayed consensus sequences now use 8 bytes per base less memory.
- NEW: Added a (prefs-file only) option to disable all disk caching for maximum performance at the expense of memory.
- NEW: The Contigs Table now lists read vs consensus mismatch percentage information.
- CHG: Some miscellaneous changes to cache support to speed up read retrieval.
- BUG: Fixed a memory-leak problem with display-time objects not being removed after closing a contig.
- BUG: Tablet now only deletes the cache files it creates, rather than every file in its cache directory.

This is the first version of Tablet to support BAM, but please note that it is just experimental support. We are currently reading BAM by parsing the entire file (as we do with all other assembly formats within Tablet). This means a large multi-GB file is read entirely by Tablet before display. Although this lets you work with BAM files, it's not ideal, so the next step is to get proper indexed BAM support working. We think we can do this without affecting how Tablet works or the number of assembly formats that it supports.

Iain

[NSTbioinformatics]

Tablet can not open Eland aligment file s_*_sorted.txt. It is pity!!!

If Tablet can handle more format of input files, it would be very nice.

By the way, is there any tool could convert s_*_sorted.txt eland alignment file to format which Tablet can read? I failed to use MAQ to convert export2maq.

[imilne]

Tablet can not open Eland aligment file s_*_sorted.txt. It is pity!!!

If Tablet can handle more format of input files, it would be very nice.

Tablet already handles ACE, AFG, SOAP, MAQ, SAM, BAM, FASTA, FASTQ, and GFF files. That's more than a lot of software out there. Of course, there's always room for more, but there's also only so many hours in the day. If you can send me (or point me in the direction) of details on the specification of the format though, we'll see what we can do about it.

By the way, is there any tool could convert s_*_sorted.txt eland alignment file to format which Tablet can read? I failed to use MAQ to convert export2maq.

I've seen messages talking about converting it into sam, so perhaps samtools can manage?

Iain

[kmcarr]

Tablet can not open Eland aligment file s_*_sorted.txt. It is pity!!!

According to the latest Illumina documentation ("CASAVA Software 1.6 User Guide", which also covers GERALD) the s_N_sorted.txt format is being deprecated so you probably shouldn't be basing your pipeline around it. Sorry to be the bearer of bad news.

[NSTbioinformatics]

Some alignment was done by eland (previous version), We are happy for the alignment and would like to view the alignment. Thus we still need to translate the format for viewing by tablet or use other tools.

SAMtool may work. I will try it.

[NSTbioinformatics]

Actually the format of s_N_sorted.txt is same to the format of s_N_export.txt file. Just s_N_sorted.txt contain the reads passed purity filitering and with a unique alignment. I think it is very useful for us. Otherwise, we have to pasor S_N_export.txt to fetch these reads.

[NSTbioinformatics]

The format of s_N_sorted.txt and s_N_export.txt is for example:
HW201 91113 5 100 1124 1381 0 1 TTTATCAAGATAATTTTTCGACTCATCAGAAATATCCGAAAGTGTTAACTTCTGCGTCATGGAAGCGATAAAACTC abbbbbbbbbabbbabbabbababaaaaa\aaaaaaaaa``aTa^a`a_a____a
`_]^^^PY_]RV^UL[BBBBB phiv2.fa 1 R 76 394 645 5246 F
*****
machinename\tRunNumber\tLane\tTile\tX coordinate of cluster\t Y coordinate of cluster\t index value\t Read number\tRead\t QualityString\tMatch chromosome\t MatchContig\tNatcgPosition\tMatch strand(F\R)\tMatch Descriptor\tSingle-Read alignment score\t paired-read alignmnet score\t partner chromosome (for second pair only)\t partner contig (for second pair)\t partner offset (for second pair)\t partner Strand (the sconed pair)\tFiltering (Y/N)\n

*****************
If imilne could make tablet read eland format, it would be very useful for us, also for users using eland for alignment

We are looking forward to the update of tablet for that.

[orcy]

Any chance for some way to change the scales on the coverage plots.

I really like this browser, and can load enough reads from a SAM to keep me happy, but there just seem to be too few "tweaks" available for the viewer.

certainly much more memory efficient than IGV in my hands.

cheers

[imilne]

Any chance for some way to change the scales on the coverage plots.

Yep, that's something we're actively working on. It might not appear for a version or two, but it's definitely on the drawing board.

Iain