A fragmented alignment method detects a phosphorylation site and a new BRC repeat in the Drosophila melanogaster BRCA2 protein, and a new HAT repeat in Utp6 from yeast [version 1; peer review: 2 approved, 1 approved with reservations]

Mutations in the BRCA2 tumor suppressor protein leave individuals susceptible to breast, ovarian and other cancers. The BRCA2 protein is a critical component of the DNA repair pathways in eukaryotes, and also plays an integral role in fostering genomic variability through meiotic recombination. Although present in many eukaryotes, as a whole the BRCA2 gene is weakly conserved. Conserved fragments of 30 amino acids (BRC repeats), which mediate interactions with the recombinase RAD51, helped detect orthologs of this protein in other organisms. The carboxy-terminal of the human BRCA2 has been shown to be phosphorylated by checkpoint kinases (Chk1/Chk2) at T3387, which regulate the sequestration of RAD51 on DNA damage. However, apart from three BRC repeats, the Drosophila melanogaster gene has not been annotated and associated with other functionally relevant sequence fragments in human BRCA2. In the current work, the carboxy-terminal phosphorylation threonine site (E=9.1e-4) and a new BRC repeat (E=17e-4) in D. melanogaster has been identified, using a fragmented alignment methodology (FRAGAL). In a similar study, FRAGAL has also identified a novel half-atetratricopeptide (HAT) motif (E=11e-4), a helical repeat motif implicated in various aspects of RNA metabolism, in Utp6 from yeast. The characteristic three aromatic residues with conserved spacing are observed in this new HAT repeat, further strengthening my claim. The reference and target sequences are sliced into overlapping fragments of equal parameterized lengths. All pairs of fragments in the reference and target proteins are aligned, and the gap penalties are adjusted to discourage gaps in the middle of the alignment. The results of the best matches are sorted based on differing criteria to aid the Open Peer Review


Introduction
The breast cancer susceptibility protein BRCA2, first identified in 1995 1 , is a critical recombinase regulator 2 that ensures genomic stability through high fidelity repair 3,4 of double stranded breaks (DSB) and prevents stalled replication forks from replicating 5 in the DNA. The primary recombinase in BRCA2 repair of DSB through homologous recombination is the RAD51 protein, belonging to the conserved RecA/RAD51 family 6 , that binds to the BRCA2 protein at various segments of ~30 amino acids (BRC repeats) 7,8 , and in the C-terminal region in most vertebrates 9,10 . Checkpoint kinases phosphorylate a serine 9 and a threonine 10 at the carboxy-terminal region of BRCA2, thereby regulating its interaction with RAD51. BRCA2 also plays a key role in fostering genomic variability through meiotic recombination 11,12 , although a different recombinase (DMC1) is implicated in this pathway in mammalian species 13 .
The BRC repeats have helped identify BRCA2 orthologs in various eukaryotic species 14 . Functional characterization of this gene in Drosophila melanogaster has demonstrated its interaction with RAD51, and a critical role in mitotic and meiotic DNA repair as well as homologous recombination 11,15 . The copy number of the BRC repeats differs considerably. The BRCA2 homolog in Ustilago maydis (a yeast like fungus) has a single BRC repeat 16 , the D. melanogaster homolog contains only three (known) repeats 14 , while there are eight repeats in the human BRCA2 gene 7 . Even among the Drosophila genus, the range of BRC repeat numbers is varied -the D. melanogaster species has only three repeats, while D. persimilis and D. pseudoobscura have up to eleven repeats 17 . RAD51 shows varying affinity for the different BRC motifs 18,19 . This difference in repeat numbers in Drosophila has raised doubts whether 'this higher repeat number is real or a genome mis-assembly artifact' 20 , and also led to speculation on the evolution of these closely related organisms 17,20 . Any such hypothesis would need to be revisited if a new BRC motif were to be identified in D. melanogaster.
In the current work, the putative threonine phosphorylation site for checkpoint kinases (Chk1/Chk2) (E=9.1e-4) and a new BRC repeat (E=17e-4) in D. melanogaster has been identified, using a fragmented technique for the pairwise alignment of two sequences (FRAGAL). The reference and target sequences are sliced into fragments of equal parameterized length X, sliding along the sequence in intervals of length Y, such that Y is less than X. Thus, the slices have overlaps. An alignment of all pairs of slices in the reference and target proteins is done using the global alignment program 'needle' 21 from the EMBOSS suite 22 . The gap penalties are adjusted to discourage gaps in the middle of the alignment. The results of the best matches are sorted based on differing criteria to aid the detection of known and putative sequences. In order to establish the generic nature of the FRAGAL methodology, the detection of a new half-a-tetratricopeptide (HAT) repeat sequence (E=11e-4) in a nucleolar RNA-associated protein (Utp6) from Saccharomyces cerevisiae is also reported. HAT is a helical repeat motif implicated in various aspects of RNA metabolism 23,24 . The characteristic three aromatic residues with a conserved spacing are observed in this new HAT repeat, further strengthening my claim 25 .
The significant conservation of the DNA repair and checkpoint pathways in flies and higher organisms 26 , the advanced genetic tools available for Drosophila, and the viability of the Drosophila BRCA2 null mutants in contrast to mammalian mutants 27 establishes Drosophila as a model organism for studying these pathways 28 . Significant divergence of key conserved sequences proves to be a serious hurdle for alignment techniques to annotate and associate the conserved sequences in the human BRCA2 to the Drosophila BRCA2 29 . Thus, a generic methodology, applicable to distantly evolutionary related proteins like BRCA2 and nucleolar RNAassociated proteins is presented. The methodology has been validated by the identification of two novel functionally relevant sites in the BRCA2 protein from D. melanogaster, and a HAT repeat in Utp6 from S. cerevisiae.

Materials and methods
The FRAGAL methodology is shown in Supplementary Figure 1. The sequences are split into fragments of X amino acids, with the starting indices sliding across the sequence length in steps of Y amino acids (SI.A.fasta and SI.B.fasta in Data Files). The score for each match is either the %similarity or a score (FRscore) that is computed as shown in Equations 1 and 2. FRscore is intended to give more weightage to identical residues in the alignment. %onlySimilarity = %similarity -%identity; (1) FRscore = 1/3 * %onlySimilarity + 2/3 * %identity; (2) One sorting criteria is to rank the matches based on the best average score, while another takes the cumulative score of a stretch of fragment matches. Stretches of fragments are stitched while ensuring the slices in the sequences are in an increasing order and nonoverlapping. The best average criteria will typically select single fragments, while the cumulative scoring criteria will bring forth longer conserved regions.
The threshold for sequence similarity for each fragment is parameterized, and set to 30% in the default mode. A large threshold will exclude more relevant matches, while a smaller threshold might include more false positives. The pairwise alignment for each fragment pair is done by a global alignment program 'needle' from the EMBOSS suite 21,22 . The parameters are set as follows -matrix=BLOSUM62, Gap penalty=25.0 and Extend penalty=0.5. The gap penalty is increased from the default value of 10 to ensure that gaps are discouraged in the middle of the alignment. Single deletions or insertions are rarely expected in conserved fragments.
The user is allowed to specify an annotation file for a given protein sequence using the uniprot accession syntax (Supplementary Figure 2). The results from FRAGAL can be filtered based on this annotation, and this provides a easier way to manually inspect and annotate corresponding segments in a query protein sequence. long and contains eight BRC repeats 7 . Further, the hBRCA2 protein is annotated for several sites phosphorylated by checkpoint kinases, which regulate its interaction with RAD51 9,10 . FRAGAL was run on the dmBRCA2 and hBRCA2 sequences. Table 1 shows the best matches obtained using two different sorting criteria -best average FRscore (see Methods) and best average %similarity -either when the match in hbrca2 is known to be conserved (Table 1A) based on an user defined input file (Supplementary Figure 2) or otherwise (Table 1B).
Detecting the threonine phosphorylation site in the carboxyterminal region of dmBRCA2. Table 1 shows a significant match (E=9.1e-4, Z-score=100) between fragment 91 of dmBRCA2 to the fragment 337 in hBRCA2, which contains the T3387 that is phosphorylated by the checkpoint kinases Chk1 and Chk2. Z-scores above a value of 8 are considered to be significant 34 . The alignment shows that the T3387 corresponds to the T926 of dmBRCA2  Detecting an additional BRC repeat in Drosophila melanogaster. The correct identification of the three BRC repeats in D. melanogaster is seen by the significant scores of the FRscore matches of A67-B152 (64), A57-B100 (60) and A75-B152 (60) ( Table 1). A significant alignment (E=17e-4, Z-score=95) between A61-B151 (35.8%similarity and 17%identity) (Figure 1c) was also observed. This sequence (634-664:LDTALKRSIESSEEMRSKASKLV-VVDTTMR) is now added to the list of sequences previously studied in the Drosophila genus 17 . The multiple sequence alignment (obtained using ClustalW 30 ) ( Figure 2a) and phylogenetic trees (obtained using PHYML 31 ) ( Figure 2b) shows that this new BRC repeat is more related to D. willistoni than other organisms in the Drosophila genus. A detailed molecular phylogeny of Drosophilid species has noted that the subgenus Sophophora is 'divided into D. willistoni and the clade of D. obscura and D. melanogaster groups', possibly indicating the source of this BRC repeat that has been conserved between D. willistoni and D. melanogaster 36 . An iterative methodology, similar to PSI-BLAST (Position-Specific Iterative Basic Local Alignment Search Tool) 37 , can be automated to generate comprehensive motifs spanning distant species. The conservation of many key residues in this sequence fragment, as shown by comparing it to the sequence logo of the Prosite BRCA2 profile (PS50138) (Figure 2c) strongly suggests that this is a putative BRC repeat. However, it must be emphasized that such repeats are to be considered putative until verified experimentally 38,39 .
Half-a-tetratricopeptide (HAT) motif HAT is a helical repeat motif implicated in various aspects of RNA metabolism and in protein-protein interactions 23,24 . These repeats are characterized by three aromatic residues with a conserved spacing 25 . A variable number of HAT repeats (9 to 12) are found in different proteins. Figure 3a shows a novel HAT repeat (E=11e-4, Z-score=116) detected in a nucleolar RNA-associated protein (Utp6) from Saccharomyces cerevisiae (Uniprot Accession:Q02354) by comparing it to HAT repeats from a human nucleolar RNAassociated protein (Uniprot Accession:Q9NYH9). Q9NYH9 has five annotated HAT repeats (121-153, 156-188, 304-335, 488-520 and 524-557), while Q02354 has three HAT repeats (87-119, 124-156 and 159-191). The new HAT sequence identified in Q02354 (SLIM-KKRTDFEHRLNSRGSSINDYIKYINYESN) is from position 30 to 62. It can be seen from the multiple sequence alignment that this sequence has the desired aromatic residues at the proper spacing, a requisite for being considered a HAT repeat (Figure 3a and b). Further, the MSA shows large variation amongst HAT sequences even A B C within the same organism ( Figure 3b). Finally, Figure 3b and c shows that certain HAT repeats are more similar to HAT repeats from other organisms than to other HAT repeats in its own sequence.

Database for aligning different pairs of BRCA2
A database (www.sanchak.com/fragal.html) which lists the results for the fragmented alignment of various proteins with BRC and HAT repeats sequences has been created. The results have been generated by varying two parameters -length of the fragments and the threshold %similarity value for a significant match in a fragment pair. As mentioned above, the results are presented in several formats -best cumulative score and best average score -where the score is either the %similarity or FRscore.

Discussion
Genetic evolution over large time spans often leaves little trace of kinship in different organisms, even when the functional roles of the genes remains conserved. A relevant example is the BRCA2 shows that certain HAT repeats are more similar to HAT repeats from other organisms than to other HAT repeats in its own sequence. In order to justify this method further, I concentrated on proteins that contain the Half-a-tetratricopeptide (HAT) repeat motifs. The HAT motif is much less ubiquitous than the related tetratricopeptide (TPR) repeat, and has been implicated in various aspects of RNA metabolism 23,24 . HAT motifs are also hypothesized to play a critical role in assembling RNA-processing complexes 25 . A recent study that combined bioinformatics, modeling and mutagenesis studies of the HAT domain used the three tandem HAT motifs in the Saccharomyces cerevisiae protein Utp6 to make inferences about the residues that confer structural and/or functional properties to the motif. In the current work, the detection of a new HAT repeat sequence (E=11e-4) in Utp6 from S. cerevisiae has been reported. This sequence has the desired aromatic residues at the proper spacing, a requisite for being considered a HAT repeat 25 . The above mentioned study would have gained further by the knowledge of this HAT repeat, a repeat that remained undetected by sequence analysis using other methods. The HAT repeats are much more varied, and thus not suitable for generating motifs (like Prosite 35 ). For example, the consensus sequence has been derived from an alignment of 742 HAT motifs from Pfam 46 and had to be manually edited since this alignment included gaps in greater than 90% of the sequences 25 . Moreover, FRAGAL detects that a particular HAT sequence in one protein is more related to HAT sequences from other species that other HAT repeats present in its own sequence. This raises interesting questions about their evolutionary history.
Existing methods for detecting functional motifs in a given protein sequence have been unable to detect these putative sites. For example, meta servers (http://myhits.isb-sib.ch/cgi-bin/motif_scan, http://www. ebi.ac.uk/Tools/pfa/iprscan/, http://www.genome.jp/tools/motif/) for detecting motifs in a protein have been unable to detect the sites identified using the FRAGAL methodology. These meta servers use one or more motif databases 35,46-49 . It is fair to mention that the FRAGAL method is much more computationally intensive than the above mentioned methods. At the same time, FRAGAL makes no assumption of any knowledge of the conserved regions (either the sequence or their position). The choice of the fragment length in FRAGAL depends on the length of repeats that is expected to be present in the protein. Since both repeats (BRC and HAT) discussed in this manuscript are around ~30 amino acid long, I have chosen a fragment length of 50. A larger fragment length might mask the similarity in the core region due to variations in the non-critical regions, whereas a smaller fragment would match irrelevant portions and thus increase false positives.
In some of the significant matches in Table 1 the fragment in hBRCA2 is not annotated to be functionally relevant -for example fragments 33 and 87 of dmBRCA2 and fragments 176 and 194 in hBRCA2, respectively. These fragments might suggest an important, yet unknown, functional relevance of that stretch of the human gene as well, since it is conserved across distant species. An excellent database for Drosophila related information is available at http://flybase.org/ 50 . A database (www.sanchak.com/fragal. html) for BRCA2 and nucleolar RNA-associated proteins from different organisms, and will be updating this on a regular basis to include more organisms and different repeats has been created. The increasing importance of Drosophila as a model system for cancer research 51 in the search for human therapeutics 52-54 can be gene which, although present in many eukaryotes, is weakly conserved 40 . The BRCA2 protein plays a major role in maintaining genomic stability, fostering genetic variability and also has other cellular functions 2,41 . Individuals with germline mutations in the BRCA2 gene are at significantly greater risk to a wide range of cancers 42,43 . This is supposed to be primarily due to the instability in chromosome structure and number induced by functional aberrations in BRCA2 44 . Conserved fragments of ~30 amino acids (BRC repeats) 7 that mediates the interaction of BRCA2 with the RAD51 recombinase 45 have been instrumental in identifying BRCA2 orthologs in other species 14,16 . The BRCA2 protein in the Drosophila genus assumes significance in this context owing to the advanced tools available for Drosophila genetics 28 , and has been functionally characterized recently 11,15 .
However, weak sequence conservation in this gene has proven to be an impediment in associating experimentally proven functionally relevant gene fragments in humans and Drosophila. The variability in the number of BRC repeats even within the Drosophila species has provided fodder for further speculation on the evolution of this gene 17,20 . The detection of a new BRC repeat would necessitate the reevaluation of such hypotheses.
Apart from the BRC repeats, RAD51 interacts with BRCA2 in the carboxy-terminal, and this interaction is modulated by checkpoint kinases 9,10 . Since the introduction of BRC repeats in the cell inhibits the formation of RAD51 nucleoprotein filaments 8 , a model has been suggested whereby RAD51 binds to both the BRC repeats and the carboxy-terminal in undamaged cells, and DNA damage triggers the release of the carboxy-terminal bound RAD51 via the phosphorylation of a threonine residue 10 .
Thus, it is noted that certain functionally significant domains are much more conserved compared to the complete protein 40 . In the current work, a methodology to annotate proteins in such 'twilight' zones 29 by fragmenting and aligning two protein sequences ( Figure 1) has been presented. The results are sorted based on differing criteria, and can be directed by a input file in case the sequences have already been annotated. This method helps in quickly honing onto conserved sites through visual inspection (Table 1 and Figure 1). The threonine phosphorylation site (E=9.1e-4) for checkpoint kinases (Chk1/Chk2) ( Figure 1) and a new BRC repeat (E=17e-4) using FRAGAL (Figure 2) has been identified. Pruning out matches which do not have a corresponding conserved sequence in hBR-CA2 helps us to select fragment 61 in dmBRCA2 as a new BRC repeat 7,14 , and fragment 91 in dmBRCA2 as the putative threonine site for phosphorylation by checkpoint kinase Chk1 and Chk2 10 . It must be noted that the sites identified remain putative until verified by experimental data, in spite of the low E-values obtained.
The multiple alignments can be used to create (for the carboxyterminal phosphorylation threonine site) or extend (for the new BRC repeat) Prosite motifs. However, the carboxy-terminal phosphorylation threonine site Prosite motif generated from the multiple alignment of sequences from Drosophila and mammals did not result in any matches in other organisms (Ustilago maydis and Caenorhabditis elegans). exploited to the hilt once the conserved mechanism is fully understood. FRAGAL presents the first step by annotating putative conserved sequence fragments in Drosophila and humans.

Competing interests
No competing interests were disclosed.

Grant information
This work was funded by the Tata Institute of Fundamental Research (Department of Atomic Energy), and the Department of Science and Technology (JC Bose Award Grant). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Figure S1. FRAGAL methodology. The BRCA2 protein sequences from Drosophila melanogaster and humans are split into fragments of parameterized length (50 in this case), at a parameterized interval (10 in this case). All pairs of fragments are aligned, and the results stitched such that there are no overlap in any given match and the order of the match is not interspersed. The alignment is done using the global alignment program 'needle' from the EMBOSS suite, and the gap penalties are set to 25 to discourage gaps in the middle of the alignment.

Department of Computer Science, University of Illinois at Urbana-Champaign, IL, USA
In this well-written manuscript, Chakraborty presents a tool for local alignment of two protein sequences that includes a fragment-chaining step. He then uses this tool to identify important putatively functional fragments in two different Drosophila proteins by comparison to the respective human ortholog. A database containing results of many more similar applications is also presented and is a nice aspect of the work.
I have the following specific comments, mostly related to the presentation, that might help the author improve the clarity of the manuscript.
The current title is rather long. The contribution of this work is mainly in the form of the FRAGAL tool, and the title could be trimmed to emphasize only that. The two example applications to finding the phosphorylation site, BRC repeat etc. are not experimentally substantiated biological claims, and may be better off being left out of the title.
Clear discussion should be provided regarding other previous work where pairs of aligned fragments are stitched together (e.g. exon chaining, see Jones & Pevzner 2004, and the chain/net approach to whole genome alignments).
Since the FRScore does not include gap penalties, I am assuming that each pair of fragments is subjected to two distinct similarity-scoring approaches; the gap-based approach when aligning that pair of fragments using 'needle' and the match/mismatch based approach when ranking the aligned pairs. This should be stated clearly, to avoid confusion. Is there a reason why the needle score was not used in place of the FRScore?
It appears that the FR score is the unweighted sum of %similarity and %identity. This should be stated explicitly.
I did not quite understand the formatting of The results of Table 1 do not aid ones understanding as to how the fragmented alignment, i.e. stitching together of fragments, helps in this case. As shown, this appears similar in form to a ranked list of matches from a standard local-aligner. Similarly, with respect to Figure 2, the author may wish to discuss why FRAGAL finds the 'melanogaster4' fragment as a BRC repeat where previous annotations (that found three repeats) failed. Was this a matter of previous methods ' missing the threshold'? (This seems unlikely given the strong E-value reported for this.) Similar clarifications for the HAT repeat finding exercise will also be helpful.

Competing Interests:
No competing interests were disclosed.

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
Author The current title is rather long. The contribution of this work is mainly in the form of the FRAGAL tool, and the title could be trimmed to emphasize only that. The two example applications to finding the phosphorylation site, BRC repeat etc. Are not experimentally substantiated biological claims, and may be better off being left out of the title.
Since this work began in search for unannotated fragments of the dmBRCA2 sequence, and is being followed in the lab actively, I think the reference to an application of FRAGAL is warranted. However, I have removed the reference to the HAT repeat and stated the fact that the phosphorylation site and the BRC repeat is putative in the title.
Clear discussion should be provided regarding other previous work where pairs of aligned fragments are stitched together...
I have discussed the spliced alignment approach to genome assembly in the discussion. However, I have noted that while these methods use graph algorithms to solve the computationally difficult problem of exon chaining, FRAGAL does the converse by finding best matches in known exon chains (i.e. protein sequences).
Since the FRScore does not include gap penalties, I am assuming that each pair of fragments is subjected to two distinct similarity-scoring approaches; the gap-based approach when aligning that pair of fragments using 'needle' and the match/mismatch based approach when ranking the aligned pairs. This should be stated clearly, to avoid confusion. Is there a reason why the needle score was not used in place of the FRScore?
The needle score includes gap penalties, which is something that is not intended for use in FRScore, as you have correctly pointed out. The idea is to direct the alignment to discourage gaps -but once the alignment is done a gap should not have a penalty. It is 'real' and therefore only the identity or similarity that matters.
It appears that the FR score is the unweighted sum of %similarity and %identity. This should be stated explicitly.
I have empirically assigned more weightage to the %identity based on the fact that we are searching for repeats, and expect higher conservation. This conservation is magnified a bit more by assigning higher weightage.
I did not quite understand the formatting of Table 1. ... I found that presenting sub-tables A and B (which I finally realized does not relate to A and B sequences) leads to more confusion than it helps.
I apologize for this confusion. The line demarcating subtables A and B was lost in the typesetting -and I missed out on detecting this error. Further, naming the subtables A and B was a poor choice of names, since the sequences were also named A and B. Finally, I agree that the second column was unnecessary, as was two subtables. I have simplified this table.

I assume something like A91-B337 refers to the starting positions of a matching fragment between sequences A and B, and the length of that fragment is not indicated in the row. Is this correct? (On reading further I realize that this interpretation is incorrect, and thenumbers in a match are arbitrary indices and not coordinates. This was not clear from the legend.)
I apologize for this oversight. This is mentioned in the web pages ... with respect to Figure 2, the author may wish to discuss why FRAGAL finds the 'melanogaster4' fragment as a BRC repeat where previous annotations (that found three repeats) failed. ... Similar clarifications for the HAT repeat finding exercise will also be helpful.
I could only make an educated guess as to why other tools failed to detect these repeats. I believe that the tools used had a 'sequential' methodology and therefore one match fixed the order of the next searches.

Satish Chikkagoudar
Pacific Northwest National Laboratory, Washington, USA The author presents an interesting technique for detecting new BRC repeats. The paper is generally well written, but needs some additional material to bolster its case. The introduction section needs more discussion of the 'state-of-the-art' in the alignment and motif detection area (especially with respect to detecting BRC repeats). The paper's argument can be made stronger by explicitly mentioning the advantages of fragmented alignment over any other recursively applied local alignment method or homology search method. Some discussion of existing methods exists in the discussion/results section, but that needs to be made available in the introduction in order to justify the need for creating a new method An explanation of the choice of parameter values needs to be given. I would like to thank you for your insightful suggestions which will help improve the manuscript. I will make the suggested changes, and incorporate them in a new version shortly.
A small clarification -you have asked for a "figure describing the FRAGAL pipeline". There is a supplementary figure S1 doing this. Do you think that this figure is insufficient, or were you suggesting that I move this to the main manuscript?
Best regards, Sandeep The extra weightage given to % identity in the score is due to the fact that one expects more sequence conservation in repeats. A figure describing the FRAGAL pipeline will be useful to visually describe the pipeline/algorithm.
I have added a pseudo code of the FRAGAL program in the main manuscript.
sequences into overlapping fragments, this method is able to discover additional motifs. The author has tested his technique on BRC repeats in Drosophila dmBRCA2 and HAT repeats in budding yeast Utp6, by comparing them to corresponding human protein sequences. The BRC repeat has been well analysed with comparisons across several Drosophila species. However the author does not provide extensive comparison of HAT repeats in Saccharomyces species. Since the sequences of several Saccharomyces sibling species and closely related fungi such as Aspergillus, Candida, etc. are known, it would be interesting to see how conserved this new HAT repeat is within the overall conservation of Utp6.
While the author establishes the advantage of FRAGAL technique, it is too early to say that this is a useful generic tool to identify known and novel motifs in protein sequences. I would request the author to run his FRAGAL code on several protein sequences with small motifs to estimate success rates and false discovery rates of his method. A supplementary table should be provided describing several sequences analysed by this method and these rates.
A minor comment, Table 1 should be simplified with the two BRCA2 protein sequences presented in two sub-tables. Please explain why certain ranks are missing in FRscore and %S.
Competing Interests: Although I am affiliated with the same institution as the author, I was not involved at any stage of manuscript preparation and did not collaborate with him at the time the review was written. I have implemented this interesting idea using proteins which have the HAT repeat from Aspergillus nidulans and Candida glabrata. This is now a Supplementary figure. However, these do not provide any further insights into the evolution of the HAT repeat, and would require sophisticated analyses beyond my expertise.

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
I would request the author to run his FRAGAL code on several protein sequences with small motifs to estimate success rates and false discovery rates of his method. A supplementary table should be provided describing several sequences analyzed by this method and these rates.
In accordance with this suggestion, I have run FRAGAL on two more motifs (BIR and TPR). However, I failed to detect any new repeats using these two motifs. These are now part of the database -http://sanchak.com/fragal.html.
A minor comment, Table 1 should be simplified with the two BRCA2 protein sequences presented in two sub-tables. Please explain why certain ranks are missing in FRscore...
I have simplified the table considerably based on the comments of another reviewer (please see below). The naming of the sub tables as A and B was confusing given that the query and target sequences were named A and B. Further; the columns for the similarity scoring has been removed. We did not ever use the similarity only score, and this was adding to the confusion. I have now clearly stated the reason for some missing ranks. I apologize for the confusing aspects of this table.
Competing Interests: No competing interests were disclosed.