background-image: url("robbie-down-wEwm7znMIDg-unsplash.jpg") background-size: cover class: left, top, inverse <p style="font-weight: 900; font-family: helvetica; font-size: 3rem; color: yellow; text-shadow: 2px 2px black;">X-chromosome EWAS</p> <p style="font-weight: 900; font-family: helvetica; font-size: 2.25rem; color: white; text-shadow: 2px 2px black;">what we found this far</p> <br> <p style="font-weight: 900; font-family: helvetica; font-size: 2.25rem; color: white; text-shadow: 2px 2px black;">Julia Romanowska</p> <br> <p style="font-weight: 900; font-family: helvetica; font-size: 1.5rem; color: white; text-shadow: 2px 2px black;">January 7, 2022</p> <p style="font-size: 12pt; font-weight: bold; right: 10px; bottom: 20px; color: white; position: absolute; text-shadow: 2px 2px black;"> Photo by <a style="color: white;" href="https://unsplash.com/@robbiedown?utm_source=unsplash&utm_medium=referral&utm_content=creditCopyText">Robbie Down</a> on <a style="color: white;" href="https://unsplash.com/?utm_source=unsplash&utm_medium=referral&utm_content=creditCopyText">Unsplash</a> </p> --- class: inverse, left, bottom # OUTLINE ##
Data ##
Methods ##
Some results ##
Discussion --- ##
Data - Boys
and girls
separately - **models** and **sample sizes**: 1. CpG ~ ART + mat_age + smoking + mat_bmi + isPrimiparous 2. CpG ~ ART + mat_age + smoking + mat_bmi + isPrimiparous + mat_meth + pat_meth 3. CpG ~ ART + mat_age + smoking + mat_bmi + isPrimiparous + GA + BW 4. CpG ~ ART + mat_age + smoking + mat_bmi + isPrimiparous + GA + BW + mat_meth + pat_meth <table style="width:100%"> <tr> <th></th> <th colspan="2" style="border-left: 1px solid black;">Model 1</th> <th colspan="2" style="border-left: 1px solid black;">Model 2</th> <th colspan="2" style="border-left: 1px solid black;">Model 3</th> <th colspan="2" style="border-left: 1px solid black;">Model 4</th> </tr> <tr style="font-weight: bold; text-align: center;"> <td></td> <td>ART</td> <td>control</td> <td>ART</td> <td>control</td> <td>ART</td> <td>control</td> <td>ART</td> <td>control</td> </tr> <tr style="text-align: center;"> <td><img src="mars-solid.svg" style="width: 20px" alt="boys"></td> <td>495</td> <td>463</td> <td>488</td> <td>456</td> <td>494</td> <td>461</td> <td>487</td> <td>454</td> </tr> <tr style="text-align: center;"> <td><img src="venus-solid.svg" style="width: 15px" alt="girls"></td> <td>446</td> <td>503</td> <td>440</td> <td>495</td> <td>446</td> <td>501</td> <td>440</td> <td>493</td> </tr> </table> ??? The main models we're discussing in the paper are Models 1 & 2, while Models 3 & 4 are mainly in the Supplementary. Models 3 & 4 expand models 1 & 2 by adjusting also for gestational age & birth weight; models 3 & 4 expand models 1 & 2, respectively, by adjusting also for parental DNAm. We saw that the most significant results were very similar in all the models, specifically, adding adjustment for GA and BW did not change the outcome. --- ##
Methods - linear regression - **BACON** - to remove bias in p-values - **DMRff** - to check for *differentially methylated regions* - **ggcorrplot** to calculate and visualize *correlation of DNAm* - **karyoploteR** & **biomaRt** - to visualize location of hits ??? Haakon Nustad did lot of work. Julia Romanowska did post-processing, bioinformatics analyses, and plotting. --- ## EWAS <img src="JRom_Xchrom_presentation_2022-01-07_files/figure-html/ewas_bacon-1.png" width="80%" style = 'position: absolute; top: 40px;' style="display: block; margin: auto;" /> ??? There were no big difference in the most significant findings across the models. In the boys-only analyses, there were mainly 3 CpGs, with 2 extra being significant in model 1. In the girls-only analyses, there were the same 6 significant CpGs in _all_ the models. --- **Significant CpGs** *significant if FDR < 0.05* <img src="JRom_Xchrom_presentation_2022-01-07_files/figure-html/ewas_signif_locations-1.png" width="65%" style = 'position: absolute; top: 80px; left: 100px' style="display: block; margin: auto;" /> ??? The significant CpGs were very similar in between the models - both effect size and FDR-adjusted p-values did not change much. There are no CpGs that are common among boys and girls, not even if we consider genomic regions. --- **DMRs** *significant if FDR < 0.01 and \#CpGs >= 3* <img src="JRom_Xchrom_presentation_2022-01-07_files/figure-html/plot_DMRs_positions-1.png" width="65%" style = 'position: absolute; top: 80px; left: 100px' style="display: block; margin: auto;" /> ??? We've found also many DMRs - interestingly more among girls than boys. Vast majority of the DMRs overlapped with promoter regions; several were co-located with CTCF (an ubiquitous TF) binding site; and only in two cases the DMR was within a gene body. --- **Co-methylation patterns** **Boys, near two most significant CpGs** <img src="coMET-like_fig_boys_model1.png" style="width: 56%; position: absolute; top: 130px; left: 150px;"> ??? And now, for something really cool! Co-methylation patterns around the significant EWAS findings. Example plot - boys-only analyses showed two CpGs that were most significant across all the models. These two are very near each other, only several bp away, so it's actually good to see that they are both picked up in our analyses. We also see that their DNAm levels are very highly correlated with each other and there is a small cluster of CpGs around those. These were picked up in the DMRff analyses. This DMR is within a short gene, _UBE2DNL_. This gene is transcribed to a protein, but its function is not known. Intriguingly, UBE2DNL protein is an enzyme that lacks a crucial cytosine in its active site, thus it is probably inactive. We know, however, that it's expressed almost only in testis. --- <!-- ## Co-methylation patterns --> **Girls, near cg13866977** <img src="coMET_girls_model1_cg13866977.png" style="width: 58%; position: absolute; top: 60px; left: 150px;"> ??? There were two equally significant results in the girls-only EWA calculations: cg25034591 and cg13866977. The last one being within a regulatory region and, at the same time, within an intron of angiomotin (AMOT) gene. Interestingly, one of the CpGs (cg13866977 was originally) annotated to a region originally defined as ‘enhancer’ in the GRCh37 genome build but this was changed to ‘promoter flanking region’ in the newer GRCh38 genome build (ensembl regulatory ID ENSR00000912938). It is not unprecedented for the definition and location of annotations to change from one genome version to another. For example, the distinction between promoter and enhancer is not always clear, especially when they have several properties and functions in common, and an update from old definitions might alter the original annotations [REF]. Hence, a more suitable annotation for cg13866977 might have been ‘transcription regulatory element’. Moreover, GeneHancer db (REF) lists this regulatory region as putative enhancer of the AMOT gene (GH ID GH0XJ112807, https://www.genecards.org/cgi-bin/carddisp.pl?gene=AMOT&keywords=AMOT#genomic_location). This regulatory region is active in only six cell types, one of which is placenta. This is based on the ensembl visualization of experimental data showing various histone markers states and DNase1 activity (http://www.ensembl.org/Homo_sapiens/Regulation/Summary?db=core;fdb=funcgen;r=X:112806973-112809972;rf=ENSR00000912938). According to The Huma Protein Atlas, AMOT protein is mainly expressed in placenta and kidney (https://www.proteinatlas.org/ENSG00000126016-AMOT/tissue), however, AMOT gene is much more transcribed in epididymis (part of the male reproductive system). Our data shows that the DNAm level at cg13866977 is almost 1 in boys (i.e., full methylation of the cytosine) and above 0.7 in majority of cases in girls (Figure_AMOT_CpG_methylation). Since DNAm signals are mainly translated to the transcription level, we looked at whether the transcription factors (TFs) that were predicted to bind to cg13866977 have preferrence to bind to the unmethylated or methylated sequence. The JASPAR and MeDReaders (http://medreader.org/browse-tf) databases show that out of seven TFs none were shown to bind to the methylated sequence, thus suggesting that the methylated state of this CpG signalises inactivation of this regulatory region. Moreover, the methylation state is higher for the girls conceived through ART than those conceived naturally (effect size 0.32, Model 1) and does not change much when we adjusted for parental DNAm at this site (effect size 0.33, Model 2). This suggests that the regulatory region within AMOT is less activated after the ART procedure in girls. --- **Zooming on DMRs...** <img src="JRom_Xchrom_presentation_2022-01-07_files/figure-html/zooming_DMR1-1.png" width="70%" style = 'position: absolute; top: 50px; left: 100px' style="display: block; margin: auto;" /> ??? One of the significant DMRs in the girls-only analyses was co-located with a promoter (ensembl ID ENSR00000249590), which according to GeneHancer DB (GH ID GH0XJ153721) is putative promoter or enhancer for nine genes (ABCD1, BCAP31, HSALNG0140785, PNCK, SLC6A8, KRT18P48, PDZD4, HSALNG0140788, and PLXNB3). Six of these genes produce proteins (ABCD1, BCAP31, PNCK, SLC6A8, PDZD4, and PLXNB3) that form a network, according to evidence from text mining and co-expression arrays (STRING DB, https://version-11-5.string-db.org/cgi/network?networkId=bWgRg6ih0nDV). Deletions or duplications of many of these genes were reported to manifest as impairments or diseases, among those autism. Right next to this DMR, the boys-only analysis showed a significant DMR co-located with a promoter (ensembl ID ENSR00002105690, GH ID GH0XJ153780) of genes SRPK3, PLXNB3, HSALNG0140793, and SSR4. Of these, only HSALNG0140793 is not a protein coding gene, however, the three other genes does not seem to have much in common. STRING DB shows no interactions (https://version-11-5.string-db.org/cgi/network?networkId=bCqD1tk1QVKv) and no common expression tissue (main RNA-expression profiles are: pancreas for SSR4 (https://www.proteinatlas.org/ENSG00000180879-SSR4/tissue), muscle for SRPK3 (https://www.proteinatlas.org/ENSG00000184343-SRPK3/tissue), and brain for PLXNB3 (https://www.proteinatlas.org/ENSG00000198753-PLXNB3/tissue)). --- background-image: url("yonatan-anugerah-3k0-K12TgLc-unsplash.jpg") background-size: cover class: top, center <p style="font-weight: 600; font-family: helvetica; font-size: 2rem; color: black;">
Questions?</p> <p style="font-size: 12pt; left: 10px; bottom: 20px; color: black; position: absolute;"> Photo by <a href="https://unsplash.com/@yonatanugerah?utm_source=unsplash&utm_medium=referral&utm_content=creditCopyText">yonatan anugerah</a> on <a style="color: black;" href="https://unsplash.com/?utm_source=unsplash&utm_medium=referral&utm_content=creditCopyText">Unsplash</a> </p> --- <!-- - why are there many more DMRs found in girls-only analyses? --> **Who wants to be on the author list?** <div style="float: right; display: inline; font-size: small;"> <p style="font-style: bold;"> LEGEND:</p> <table style="width: 250px; right: 50px;"> <tr> <th>symbol</th> <th style="border-left: 1px solid black;">responsibilities</th> </tr> <tr> <td style="text-align: center;">
</td><td>playing with data</td> </tr> <tr> <td style="text-align: center;">
</td><td>contribute to discussion; biological expertise</td> </tr> <tr> <td style="text-align: center;">
</td><td>checking literature, making text shine</td> </tr> <tr> <td style="text-align: center;">
</td><td>supervision</td> </tr> <tr> <td style="text-align: center;">
</td><td>suddenly appearing and asking <i>the most to-the-point question</i></td> </tr> <tr> <td style="text-align: center;">
</td><td>friendly reviewing the bioinf. approach and implementation</td> </tr> </table> </div> currently: -
*Julia Romanowska*, -
*Haakon E. Nustad*, -
*Christian M. Page*, -
*William R.P. Denault*, -
*Jon Bohlin*, -
*Yunsung Lee*, -
*Maria C. Magnus*, -
*Håkon K. Gjessing*, -
*Robert Lyle*, -
*Per Magnus*, -
*Siri E. Håberg*, -
*Astanand Jugessur*