Dear Yuan-Yuan Cheng,

Please use the SITECON - the only computational program fully proved to assess, evaluate and predict functionally active transcription sites in regulatoru region of mammalian and viral genomes. Please see attached our paper using SITECON for quantification of transction DRE sites in genes encoding for the AhR components. 

Best wishes,

Ilya

The MEME suite available at (http://meme.nbcr.net/meme/) has an algorithm for de novo motif discovery. You should be wary of the settings you use and which sequences you provide to avoid false positives/negatives- but any motifs you find that are enriched will need to be experimentally tested anyways. Maybe conservation of X between bacterium and a subsequent literature search can also help you reach your goal.

Cheers

Dear Yuan-Yuan Cheng and Ryan M Boileau,

With all due respect, no way the MEME suite is even close to the SITECON efficacy. Namely, we used MEME suite while de novo DRE core sequences discovery in 5’-flanking region of the HSV-1 genes. The MEME’ DREs were recognized with a MS score of 0.8357, that was even lower than accepted threshold (0.85) for any study, and much lower than recognition threshold of 0.95 in current study.  Instead,  with SITECON, a total of 13 bona fide DREs, all including the substitution intolerant core sequence (5’-GCGTG-3’) and adjacent variable sequences, were used as a training sample to detect conservative conformational and physico-chemical properties for the site with DRE, and to construct the recognition rules and determine conformational similarity score threshold to rank identified DREs. We used the approach utilized for mammalian DREs, but the recognition threshold of 0.95, for which type I error equal to zero, i.e., the condition when SITECON employed conformational similarity score allowing recognize all sites in the training sample. Earlier, an efficacy of SITECON in detection of functionally active DREs was tested using sequences of genes encoding human CYP1A1 and CYP1B1, which was also proven experimentally. It was demonstrated that at the recognition threshold of 0.95, SITECON efficiently identified all tested functionally active DREs. In fact, promoter DREs within genes encoding AhR, HSP90, XAP2, and macrophage pro-inflammatory factors were detected by SITECON with high scores of 0.974-0.9879.

Best wishes,

Ilya

Ilya,

Hope I am understanding this correctly-did you not have to provide the 13 bonafide DRE's to construct recognition rules for SITECON? The power in SITECON seems to be using previous collected data to conduct an informed search in conjunction with some extra tweaks to the algorithm about the biochemistry of the target/protein. For a de novo search of enriched motifs without a priori knowledge of target sequences it seems as though SITECON may lose some of its efficacy. I am not saying MEME is better, but in this situation I suppose SITECON may lose some of its advantages.

Thank you for your consideration and insight, I might be using SITECON in the future.

If you have access to Genomatix Software Suite in your institute, try it. It is one of the best for predicting transcription factor binding sites and promoter regions. 

You can use FootPrinter or MicroFootPrinter (if you are working on prokaryotic genomes) and apply it to the set of promoters of co-expressed genes. MEME allows you to identify overrepresented motifs (candidates regulatory elements) on your promoters, but these motifs could differ by several nucleotides (not really conserved).

Regarding SITECON, I didn't use it  before. I hope that this can help you.

Hello all,

Dr. Yuan-Yuan Cheng wrote, "I found a bacterial transcription factor X proved by EMSA. Transcriptome showed that some genes' expression are sigenificantly altered by X". So, "a priori knowledge of target sequences" is, in fact, in easy reach of every student. Immediately after that baby step, the SITECON might be (rather must be) used.

Sincerely,

Ilya

Thanks everybody. As I understand, if I'm right, SITECON needs some confirmed sequences to train it before predicting unidentified sequences. While MEME can predict conserved binding site de novo, however it might loose some accuracy. I will try.