-What is Exalign?
Exalign is a tool that performs alignments of gene exonic structures. In particular it works by aligning arrays representing the exon lengths of genes. While the aligning algorithms used by Exalign are the well known Smith-Waterman and Needleman-Wunsch ones (with a suitable scoring function), the program performs a post-processing step that tries to identify intron gain/loss events (see sections about exon merging). For further explanation of how the alignment algorithm Exalign is based on works please refer to the article.
-What Exalign can do for me?
Given your favourite gene (from now on: the query gene) Exalign can align its exonic structure against the structure of another gene of your choice. Otherwise Exalign can exhaustively align the structure of your query gene to the ones of all the genes annotated for many organisms, ranking the results according to alignment scores and relative expected values (E-values) computed in a BLAST-like fashion (see article).
-Are the results of Exalign coming from nucleotidic or peptidic sequences alignment?
No, Exalign performs only alignments of exonic structures or, in other words, alignments of arrays of exon lengths. Whenever possible, Exalign will also run a "Blast2Seq" for every pair of proteins and transcripts of genes whose structures are aligned, but this is only done to give more completeness to the output, as Exalign does not use the sequences alignment in any way to rank the results obtained by structure similarity.
-How to use Exalign?
Exalign comes with an user friendly web interface. Let's explain the simple steps needed to use it.
1- Fill the query field. The first thing to do is to fill in the "Query RefSeq ID" field with the RefSeq ID of one of the RefSeq transcripts of your query gene (most genes only have one RefSeq transcript). You can use any RefSeq transcript ID from Human, Mouse, Rat, Cow, Opossum, Chickem, Xenopus, Zebrafish, Fugu, Tetraodon, Drosophila, C. elegnas, Sea Urchin and C. intestinalis. Other than the NM_ RefSeq IDs it's possible to use the XM_ ones (the XM_ is used for predicted transcripts) of the same organisms. It's even possible to use IDs of "custom genes", that are genes you can design if your favourite gene is not present in our database. See below for further details on how to design your custom gene(s).
2- Pick an alignment type. You can choose between three different alignment algorithm types. The "Local" one uses the Smith-Waterman algorithm, the "Global" one uses the Needleman-Wunsch algorithm, while the "Glocal" type is an hybrid of the other two algorithms: it performs a global alignment where the gap penalties are not counted for gaps located at the extremities of the alignment.
3- Decide what you want to do. If you wish to align your query gene structure against the one of another gene you can fill the "Target RefSeq ID" field. In this field you can put every ID you can use in the "Query" field (see 1) with the exception of the "Custom Genes" IDs. As an alternative you can decide to search across the whole set of gene structures of many organism in a BLAST-like way ("database search" mode), looking for the genes with the most similar structure to the one of the query gene. Notice that the search will be performed only on the "NM_" RefSeq transcripts unless the "Include XM genes" option in the "Advanced Options" panel is selected.
4- Tuning the Advanced Options. In most cases you do not really need to tweak the Advanced Options, so feel free to skip this paragraph, but if you like to do it just click on the "Toggle Advanced Options" link to activate the advanced options panel. In the first two fields you can change the "Opening gap penalty" and the "Continuing gap penalty". The default values are 0, that does not mean that there are not gap penalties but that the program dynamically chooses the penalties according to how many exons are present in the query gene. If you want to set your own gap penalties values please remember that both numbers must be positive and that the "Opening gap penalty" value has to be greater than the "Continuing gap penalty" one. Please note that changing those values could heavily affect the E-value statistic accuracy. Then there is the "Include XM genes" option: checking this box the exhaustive search is done not only among the "NM_" RefSeq transcripts of the selected organism(s) but also among the "XM_" (predicted) RefSeq transcripts when available. The "5' Skipped Exons" and the "3' Skipped Exons" fields will change the number of external exons that will not be considered during the alignment process. The default value is 1 for both fields, meaning that the first and the last exon of every gene will not be aligned. Those fields could be useful if your query gene has more than one untranslated exon at the beginning (so when you could wish to increase the number of ignored exons at the 5' ) or very short UTRs (in this case you could want to set to 0 one or both of those fields). The "Display top" field simply affects the number of results you will have in your output when you do an exhaustive search across a genome. Please note that you will obtain the chosen number of results for every selected genome. Increasing this number might be of use when you expect a gene to have several paralogs, in order to have a more comprehensive view of the structural relationships among members of the same gene family. The "Only Results with Blast hit" checkbox, if selected, will allow only the structure alignments with at least one Blast2Seq hit in their protein alignment to be shown in the output. Please note that if this checkbox is selected and there are no alignments with at least one "blastp" hit then the highest ranking structure alignment will be nevertheless shown. Finally the "Blastn word size" and the "Blastp word size" will affect the corresponding parameters of the "Blast2Seq" algorithm. As already said Exalign only use "Blast2Seq" to integrate its output with more informations, so changing those parameters has no effect on the structure alignment.
5- Click the "SUBMIT QUERY" button. A "please wait" page will be displayed. Usually, with average server load conditions, the elaboration takes no more than a pair of minutes (something more if you select the exhaustive search across many genomes). Please wait without reloading the web page since this may cause some visualization problem and you might not be able to see your results.
-How to read the results.
The output is summarized in a page where you can have a first overview of the results: you can anyway click on every single alignment and look at it in the details.
1- Exalign Results page. If you chose to align only two gene structures you naturally have only one result, instead if you chose the exhaustive search across one or more genomes the results are ranked by descending score and only the top five (unless you have changed it in the "Advanced Options" panel) highest scoring alignments for every selected genome are displayed. Every alignment is contained in a white box where you can find the names (or the IDs) of the aligned genes and the organism to which they belong, the score and the E-value of the alignment and an image representing the alignment itself. When available, under the image you will find two links that bring you to the results of "Blast2Seq" for the proteins and the transcripts alignments. In the image representing the alignment every gene is represented by a series of coloured rectangles and a series of coloured lines. The rectangles represent the exons while the lines represent the reading frame of every coding exon.
2- Color code for the exons.
-Dark green. There is a perfect match between the lengths of two aligned exons in that position of the alignment.
-Middle green. There is a nearly exact match (delta <= 12) between two aligned exons sharing the same reading frame.
-Light green. There is a nearly exact match (delta <= 12) between two aligned exons that do not share the same reading frame.
-Light red. The length of the two aligned exons is very different (delta > 12) but the exons share the same reading frame.
-Red. The length of the two aligned exons is very different (delta > 12) and the two exons do not share the same reading frame.
-Yellow. The exon in that position is aligned with a gap.
-Horizontal black line. In this position there is a gap in the alignment.
-Purple. This exon has not been considered during the alignment (usually they are the two external exons of each gene) and is displayed only for completeness.
-Diagonal blue line(s). This line(s) appears inside an exon when the exon itself is the result of the "merging" of two (or three) shorter exons. When this line(s) show up it is possible that what we are looking at is an intron gain/loss event between the two aligned genes (see also "How does merging work" section).
-Black circle in the center of an exon. The black circle means that there is a "Blast2Seq" hit (even if it is partial) for the proteins alignment in the region of the protein encoded by this exon.
Note: you will not see purple exons when you use the local alignment algorithm. In fact only the aligned exons are shown when you use local alignment.
3- Color code for the reading frames.
-Dark blue. Frame 0.
-Middle blue. Frame 1.
-Light blue. Frame 2.
4- The meaning of the E-value. The E-value is a statistical significance measure of an alignment, and it represents the expected number of alignments with the same score or higher that can be obtained by chance when searching a database of the same size of the one used by Exalign using a gene with the same number of exons of the query gene. See the main Exalign paper for further details on how E-values are computed. They should be read as BLAST E-values are read, in other words, those approaching one are likely to be associated to non-significant/random alignments. However, short genes (e.g. with four or five exons) by definition cannot yield alignments with low E-values (as nucleotide/protein sequences of a few residues with BLAST). In this case, and also in the more general case of "borderline" (according to E-values) alignments, a good idea is to check whether Blast2Seq found also significant similarities in the protein/nucleotide sequences of the genes. The latter are usually a clear indication of relatedness between two or more genes.
5- Alignment details page. Clicking one of the images representing one of the alignments you open the alignment details page for the corresponging alignment. In this page the color code for the exons and the reading frames are basically the same of the summary page with few differences. The first difference is that the coloured rectangles now contain a number that correspond to the length of the exon they represent; then, exons with a blast hit are no more represented by a black circle but by a black box surrounding that number. Another difference is that in non coding exons, or in the non coding part of an exon, the color is now patterned. Exons resulting from a merge of two or three other exons display the lengths of the original exons too. If the alignment algorithm used is the local one you find in every exon also a number between brackets that represents the position of that exon in the gene; moreover you find two more images that show you which exons of the two genes have been used in the local alignment (the green ones) and which not (the yellow ones).
-How does the merging step work?
Merged exons are an indicator of a possible intron gain/loss detectable by comparing two homologous genes (see article). As explained in the article, Exalign has two different and well separated "exons merging steps".
The first one searches, for both the genes of an alignment, pairs (or triplets) of exon lengths that sum exactly to the length of an exon of the other gene in the alignment. When the algorithm finds such a couple (or triplet) of exons it checks if merging them together in a single exon will yield a better (higher scoring) alignment and, if so, it keeps the new alignment with the merged exons. When you use the "database search" mode, since the merging step could be very computationally intensive, merging itself is limited to the top (highest scoring) "x" alignments, where "x" is proportional to how many results you decided to display in the advanced options menu (default is five). Notice that this could cause, in rare occasions, some apparently strange behaviour by Exalign: it is in fact possible that by increasing the number of results displayed not only new results will be displayed but also the ranking of the results themselves might change. This happens because more alignments are now processed by the first merging step and if one or more of those alignments now contain merged exons their score will be greater than before, thus affecting the results order itself.
The second "exons merging step" is performed only in post-processing and it is limited to the "one vs. one" alignment mode (it will not be executed with the "database search" mode) . Given an alignment it searches for couples of exons, surrounded by exact matches, that are aligned with a mismatch and a gap. Once such a configuration is found Exalign checks if the sum of the lengths of the two exons is similar (delta <= 21 and divisible by 3) to the length of the exon aligned with them. If this is the case the two exons are merged together and the gap in the alignment is removed. Since this is a post process step the score of the new alignment is not recalculated and it remains the same of the original alignment.
-Using the gene design tool.
If your favourite gene is not available in our databases but you would like to use it with Exalign you can design its structure. The first thing to do is to click on the "Or design your own gene" link in the Exalign home page. Now you have to: choose a name for your gene (the first field), the number of exons it has (second field) and select the organism to which it belongs. Once you click on the "SUBMIT" button a new page is loaded where you have to write the length of every exon composing your gene and the start and end positions of the coding sequence (relative to the 5' end of the transcript). Finally, you can also input transcript and a protein sequences for your gene. Now click the "SUBMIT" button and write down the ID of the new gene that usally is composed by the string "CUSTOM_" followed by the name you chose. From now on you can use this ID in the "Query" field every time you run Exalign.