Cohesin Loss Eliminates All Loop Domains

May combine two claims.

Whereas loop domains disappear entirely after cohesin loss, compartmentalization is preserved.

“Whereas loop domains disappear entirely after cohesin loss, compartmentalization is preserved (Fig. 3A).”

May combine two claims.

After auxin withdrawal, APA signal quickly increased, largely recovering by 1 hour.

“After auxin was withdrawn, the APA signal quickly increased, largely recovering by 1 hour (Fig. 2B, S2B).”

May combine two claims.

Faster recovery was strongly associated with high NIPBL binding between loop anchors, enrichment of promoter and enhancer elements, and activating histone marks H3K36me3 and H4K16Ac.

“Faster recovery was strongly associated with high levels of NIPBL binding between the loop anchors, as well as enrichment of promoter and enhancer elements, and of activating histone marks like H3K36me3 and H4K16Ac.”

May combine two claims.

After auxin treatment, algorithms found only 9 loop domains, all false positives.

“Moreover, the algorithms found only 9 loop domains after auxin treatment. Upon close inspection, all were found to be false positives (see Methods).”

Classified as probably not a substantial claim.

Individual recovery curves were calculated for 2038 of the 2140 loop domains observed in untreated cells.

“we were able to calculate individual recovery curves for 2038 of the 2140 loop domains observed in the untreated cells (Fig. 2C, see Methods).”

May combine two claims.

In the time-course, APA signal was initially strong, gone by 40 minutes after auxin treatment, and remained absent throughout treatment.

“The APA signal was initially strong, but was gone by 40 minutes after treatment, and remained absent throughout the rest of the treatment period (Fig. 2B, S2B).”

May combine two claims.

Strong effects were infrequent: 64 genes (0.5%) showed a 2-fold change, and 2 genes showed a 5-fold change.

“Strong effects were infrequent: 64 genes (0.5%) showed a 2-fold change, and 2 genes showed a 5-fold change (Fig. 7B).”

May combine two claims.

Superloops tend to occur simultaneously, forming hubs containing 3 or more loci.

“We also found that the superloops tend to occur simultaneously, forming hubs containing 3 or more loci.”

May combine two claims.

In the presence of cohesin, the majority of these superenhancers were located in the interior of cohesin-associated loops (13 of 19 cases) and the long-distance cohesin-independent links were much weaker.

“By contrast, in the presence of cohesin, the majority of these superenhancers were located in the interior of cohesin-associated loop (in 13 of 19 cases) and the long-distance cohesin-independent links were much weaker.”

May combine two claims.

In untreated cells, 1,030 loops not associated with contact domains were annotated; after cohesin loss, only 72 were annotated.

“Whereas 1,030 such loops were annotated in untreated cells, only 72 were annotated after cohesin loss.”

May combine two claims.

For compartment boundaries lying in the interior of a loop domain in untreated cells, the correlation in genome-wide contact pattern on opposite sides showed a much greater decrease in treated vs untreated cells, meaning the plaid pattern became stronger in the absence of cohesin.

“In the former case, the correlation in the genome-wide contact pattern on opposite sides of compartment boundaries showed a much greater decrease in treated vs. untreated cells—that is, the plaid pattern across the genome became much stronger in the absence of cohesin (Fig. 3C–E, Data S1,III).”

May combine two claims.

They identified 57 million unique triples, 32 million unique quadruples, and 130 thousand quintuples.

“We identified 57 million unique “triples,” which bring together 3 loci; 32 million unique quadruples, and 130 thousand quintuples (Fig. 5B, see Methods).”

May combine two claims.

Histone tail interactions overcame constraints on the chromatin chain, leading either to focal contacts between short chromatin intervals or to global condensation, depending on the initial configuration and epigenetic state of the fiber.

“We found that histone tail interactions overcame constraints on the chromatin chain, leading either to focal contacts between short chromatin intervals or to global condensation, depending on the initial configuration and epigenetic state of the fiber (Fig S6C–E).”

Probably not a substantial claim: may not pass S3 (load-bearing).

In total, 203 interchromosomal links were identified.

“In total, we identified 203 such interchromosomal links.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

A 4.5 observed/expected enrichment cutoff eliminated rearrangement peaks, removing nearly equal numbers from untreated and treated maps (277 vs 269).

“Empirically, this max threshold removed peak annotations due to rearrangements; notably, nearly the same number of annotated peaks were removed from the untreated and the treated annotations, 277 and 269 respectively.”

May combine two claims.

Domain calls numbered 9,845 in untreated maps and 2,090 in treated maps.

“We annotated 9,845 domains in our untreated maps and 2,090 domains in our treated maps.”

May combine two claims.

Loops and links between superenhancers increase in strength rapidly after cohesin loss and reach a plateau within hours.

“Notably, loops and links between superenhancers increase in strength rapidly following cohesin loss, reaching a plateau within hours.”

Classified as probably not a substantial claim.

Loop calls numbered 3,170 in untreated maps and 81 in treated maps.

“In the end, we annotated 3,170 loops in our untreated maps and 81 loops in our treated maps.”

May combine two claims.

Cohesin loss did not significantly change total transcription globally, as matched pairs had similar Drosophila spike-in read fractions.

“We observed similar fractions of reads mapping to the Drosophila spike-in in the matched pairs of degron-expressing and control replicate experiments, indicating that there are not significant global changes in the total amount of transcription upon cohesin loss.”

May combine two claims.

In this study's auxin-treated maps, APA signal is completely lost and no visible focal enrichment remains, even at low resolution.

“By contrast, our maps after auxin treatment show complete loss of APA signal and no visible focal enrichment, even when APA is performed on low resolution data sets (Fig. 2B, S2C).”

May combine two claims.

Top 10% by H3K27Ac density (>=54.4 peaks/Mb) showed 38% recovery by 40 min, bottom 10% (<=2.76 peaks/Mb) showed 8%.

“Similarly, the top 10% of loop domains by density of H3K27Ac peaks (>=54.4 peaks/Mb) showed 38% recovery by 40 minutes, while the bottom 10% of loop domains by density of H3K27Ac (<=2.76 peaks/Mb) showed only 8% recovery by 40 minutes (Fig. S2D).”

Probably not a substantial claim: may not pass S3 (load-bearing).

Mean correlation of first principal component eigenvector between untreated and auxin-treated maps across chromosomes was 0.968.

“We then calculated the correlation of the eigenvector for each chromosome between untreated and treated maps. The mean correlation was 0.968.”

May combine two claims.

Among the 9 treated calls, 8 came from rearrangements and 1 from compartmentalization, all false positives.

“Of the 9, 8 were false positives due to rearrangements in HCT-116 cells and one was a false positive due to extensive compartmentalization that was mistakenly annotated as a loop by HiCCUPS.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

They identified 4,325 boundaries in untreated cells for median compartment size 425 kb and 4,424 boundaries in treated cells for median 475 kb.

“We identified 4,325 boundaries in untreated cells for a median compartment size of 425 kb and 4,424 boundaries in treated cells for a median compartment size of 475 kb.”

May combine two claims.

All-domain ADA scores drop sharply in the first 60 minutes after auxin but level off above 1, suggesting leftover domain signal.

“While the ADA scores for all domains show a sharp decline in the first 60 minutes after auxin treatment, they plateau above 1, indicating some residual domain signal from al domains.”

May combine two claims.

Active promoter and enhancer positions remain mostly unchanged after cohesin loss, with H3K4me3, H3K4me1, and H3K27Ac peak overlaps of 90%, 85%, and 81%.

“High overlap (90%, 85% and 81% respectively) was seen for all three marks, indicating that the positions of active promoters and enhancers are largely unchanged after cohesin loss (Fig. S1D).”

May combine two claims.

Re-analysis of three prior studies showed positive APA scores and visible focal enrichment in all Hi-C experiments before and after cohesin or CTCF depletion.

“We observed positive APA scores (>1) and visible focal enrichment in all experiments generated in previous studies, before and after cohesin or CTCF depletion (Fig. S2C).”

The quote only partly states this claim.

Restricting to loop domains with no NIPBL or H3K27Ac binding within 50kb of either anchor, similar enrichment patterns for activating factors, particularly strong NIPBL sites and superenhancers, appeared at fast vs slow recovering domains.

“We observed similar enrichments of activating factors, and especially strong NIPBL sites and superenhancers at the loop domains that recovered more quickly vs those that recovered more slowly.”

May combine two claims.

Top 10% of loop domains by NIPBL peak density (>=24 peaks/Mb) showed 38% recovery by 40 min, while bottom 10% (0 peaks/Mb) showed 11%.

“We found that the top 10% of loop domains by density of NIPBL peaks (>=24 peaks/Mb) showed 38% recovery by 40 minutes, while the bottom 10% of loop domains by density of NIPBL peaks (0 peaks/Mb) showed only 11% recovery by 40 minutes.”

May combine two claims.

Upon auxin removal, loop domains re-form quickly, showing strong signal by 60 minutes.

“After withdrawal of auxin, loop domains form similarly quickly, with strong loop domain signal by 60 minutes after withdrawal.”

May combine two claims.

Visual inspection classified 66 of the 81 as false positives; 55 of those came from assembly problems, repeats, or rearrangements.

“When we visually examined these loops, we found that 66 were false positives, with 55 of the false positives due to assembly issues, issues with repetitive elements or structural rearrangements.”

The quote only partly states this claim.

For H3K27me3 transitions spanned by loop domains, stronger increase: 0.01 increase untreated vs 0.03 dip treated.

“there was a stronger increase in compartmentalization strength at transitions spanned by loop domains (0.01 increase in sliding correlation in untreated vs. 0.03 dip in treated).”

May combine two claims.

For boundaries located at loop anchors, the post-treatment gain in compartment strength was smaller, with sliding correlation dropping by 0.35 untreated versus 0.53 treated.

“In contrast, when we identified 389 compartment boundaries in treated cells that were positioned at loop domain anchors annotated in untreated cells (within 25kb), we found that there a much more modest increase in compartment strength after treatment: 0.35 decrease in the sliding correlation score in untreated cells vs. 0.53 in treated cells (Fig. 3E).”

The quote only partly states this claim.

For H3K27me3 transitions at loop domain boundaries, little change: 0.29 dip untreated vs 0.33 treated.

“However, while there was very little change in the compartmentalization strength at transitions at loop domain boundaries (0.29 dip in sliding correlation in untreated vs. 0.33 in treated),”

May combine two claims.

Superenhancers were strongly enriched at loop anchors from the 88 loop automated list; top 100 superenhancers were 47-fold enriched (30/115 anchors).

“We also found that superenhancers were strongly enriched at loop anchors generated from the 88 loop list automatedly called with low resolution HiCCUPS; the top 100 superenhancers were 47-fold enriched (present at 30/115 loop anchors).”

May combine two claims.

Of these 88 loops, 46 were true positives and 42 false positives (22 repetitive regions, 15 other structure).

“After visual examination, we found that 46 of these loops corresponded to true positives while the other 42 were false positives (22 were due to issues with repetitive regions and 15 were due to other forms of structure in the contact map, for instance interactions between broad compartment intervals).”

May combine two claims.

For H3K27Ac transitions spanned by loops, sliding correlation decrease 0.41 untreated vs 0.50 treated; at loop anchors, 0.10 untreated vs 0.26 treated.

“We saw an 0.41 decrease in the sliding correlation score in untreated cells vs. 0.50 decrease in treated cells for H3K27Ac transitions spanned by loops, and an 0.10 decrease in the sliding correlation score in untreated cells versus an 0.26 decrease in treated cells for H3K27Ac transitions at loop anchors (Fig S3B).”

Classified as probably not a substantial claim. The quote only partly states this claim.

Using modified HiCCUPS at 50 and 100kb with specified parameters, they annotated 88 loops.

“This annotation yielded 88 loops.”

• cites Rao et al., 2014 sourceNot checked yet

  Not checked in the free run.

• cites Darrow et al., 2016 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Automated, uncurated lists of cohesin-independent loops gave similar results: reduced CTCF binding and no CTCF orientation preference.

“We also performed the analyses listed above on automated lists of cohesin-independent loops without any manual curation. We found that the results showing a lack of CTCF binding at cohesin-independent loop anchors and a lack of CTCF orientation preference were similar (Fig. S4A).”

May combine two claims.

They found 391 untreated-cell H3K27me3 transitions crossed by loops and at least 100 kb from either anchor, plus 469 transitions at loop anchors.

“Similar results were seen for H3K27me3 transitions spanned by all loops: we identified 391 H3K27me3 transitions in untreated cells that were spanned by loops and at least 100kb away from either loop anchor, and we identified 469 H3K27me3 transitions in untreated cells that were positioned at loop anchors.”

May combine two claims.

They found 593 treated-cell compartment boundaries crossed by loops and at least 100 kb from either anchor, plus 503 boundaries sitting at loop anchors.

“we identified 593 compartment boundaries in treated cells that were spanned by loops and at least 100kb away from either loop anchor, and we identified 503 compartment boundaries in treated cells that were positioned at loop anchors.”

Probably not a substantial claim: may not pass S3 (load-bearing).

They identified 64 loop anchor loci from the 61 loops by merging adjacent loci and expanding to 100kb.

“We first identified the loop anchors contributing to the cohesin-independent loops. We merged all adjacent loci involved in one of the 61 loops annotated above. We then expanded all loop anchor loci to be 100kb in size, yielding a list of 64 loop anchor loci.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

They found 426 untreated-cell H3K27Ac transitions crossed by loops and at least 100 kb from either anchor, plus 381 transitions at loop anchors.

“Similar results were seen for H3K27Ac transitions spanned by all loops: we identified 426 H3K27Ac transitions in untreated cells that were spanned by loops and at least 100kb away from either loop anchor, and we identified 381 H3K27Ac transitions in untreated cells that were positioned at loop anchors.”

May combine two claims.

CTCF binding occurred at 90% of cohesin-associated loop anchors but only 20% of cohesin-independent ones.

“We found that while 90% of cohesin-associated loop anchors were associated with CTCF binding, only 20% of cohesin-independent loop anchors were associated with CTCF binding.”

May combine two claims.

In treated cells, compartment domains had a median corner score at the 89th percentile of random scores, whereas non-compartment domains had one at the 72nd percentile.

“We then analyzed the corner scores for each of these sets of domains separately in treated cells and found that while the median score for compartment domains was 89th percentile of the random corner scores, the median score for non-compartment domains was only 72nd percentile of the random corner scores.”

May combine two claims.

Among cohesin-associated anchors, 95% of unique CTCF motifs faced the loop interior, consistent with the convergent rule, whereas cohesin-independent loops showed no such orientation bias (56% inward).

“More over, while 95% of unique CTCF motifs in cohesin-associated loop anchors pointed towards the interior of the loop (consistent with the convergent rule), the unique CTCF motifs in cohesin-independent loops did not exhibit any such bias (56% pointing towards the interior of the loop) (Fig. 4C).”

Classified as probably not a substantial claim. May combine two claims.

They labeled 974 contact domains as compartment domains when their edges coincided with a compartment interval edge within 25 kb, and 410 as high-confidence non-compartment domains when their edges lay more than 100 kb from any compartment boundary.

“To test this, we identified 974 contact domains whose boundaries overlapped a compartment interval boundary (within 25 kb), i.e. compartment domains. Additionally, we identified 410 contact domains whose boundaries were not within 100 kb of a compartment boundary even after using a relaxed threshold for identifying compartment boundaries, i.e. a high confidence set of non-compartment domains.”

May combine two claims.

For boundaries spanned by loops, sliding correlation decrease was 0.11 untreated vs 0.37 treated; for boundaries at loop anchors, 0.38 untreated vs 0.54 treated.

“We saw an 0.11 decrease in the sliding correlation score in untreated cells vs. 0.37 decrease in treated cells for compartment boundaries spanned by loops, and an 0.38 decrease in the sliding correlation score in untreated cells versus an 0.54 decrease in treated cells for compartment boundaries at loop anchors (Fig S3D).”

Classified as probably not a substantial claim. The quote only partly states this claim.

Using default HiCCUPS parameters at 5,10,25kb with <4.5-fold enrichment filter, they annotated 81 loops in treated RAD21-mAC cells.

“Using this procedure, we annotated 81 loops in treated RAD21-mAC cells.”

• cites Durand et al., 2016bNot checked yet

  Not checked in the free run.

• cites Rao et al., 2014 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

For boundaries inside loop domains, compartment strength rose sharply after loop loss, with the sliding correlation score dropping by 0.10 in untreated cells versus 0.31 in treated cells.

“We observed that the boundaries contained within loop domains showed a strong increased in compartment strength (larger dip in the sliding correlation score) after the elimination of loop domains: 0.10 decrease in the sliding correlation score in untreated cells vs. 0.31 in treated cells.”

May combine two claims.

For H3K27me3 transitions spanned by loops, sliding correlation decrease 0.27 untreated vs 0.31 treated; at loop anchors, 0.03 untreated vs 0.12 treated.

“We saw an 0.27 decrease in the sliding correlation score in untreated cells vs. 0.31 decrease in treated cells for H3K27me3 transitions spanned by loops, and an 0.03 decrease in the sliding correlation score in untreated cells versus an 0.12 decrease in treated cells for H3K27me3 transitions at loop anchors (Fig S3C).”

May combine two claims.

In oligonucleosome simulations of alternating wild-type and folded-tail nucleosomes, demarcated zones emerged: wild-type nucleosomes segregated from folded-tail nucleosomes and tended to associate with each other, while folded-tail nucleosomes segregated separately.

“As seen in the contact maps and corresponding images in Fig. S6E, demarcated zones emerge in the ensemble of the alternating constructs: the wt nucleosomes segregate from the other nucloeosmes as well tend to associate with one another, while the folded-tail nucleosomes segregate separately.”

May combine two claims.

At 100-kb resolution, APA showed that cohesin-independent links, both intra- and interchromosomal, were weak before auxin and became about twice as strong after treatment.

“APA analysis clearly demonstrated that while cohesin-independent links (both intra and interchromosomal) were weakly present before auxin treatment, they were ~2-fold strengthened after auxin-treatment (Fig S4C,D).”

May combine two claims.

In the auxin-treated data set, 11 contacts fell in the center voxel of the 3D-APA sub-tensor, while none did in the untreated data set.

“Using our triple dataset as well as the Poisson projection of our quadruples, we identified 11 contacts that fell within the center voxel of the 3D-APA sub-tensor in our auxin-treated dataset, where as we saw none in our untreated dataset.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Results with only H4 tails folded, mimicking H4K16Ac, were very similar to the all-folded tails case.

“Results with only H4 tails folded (mimicking H4K16Ac) are very similar to the case of all-folded tails (data not shown).”

Probably not a substantial claim: may not pass S3 (load-bearing).

Among expressed genes, 64 (0.5%) showed a 2-fold change.

“Stronger effects were seen, but less frequently: 64 genes (0.5%) showed a 2-fold change, and 2 genes showed a 5-fold change (Fig. 7B).”

May combine two claims.

APA scores for cohesin-independent links rose rapidly after auxin treatment and fell rapidly after auxin withdrawal.

“APA scores for cohesin-independent links rapidly increased upon auxin treatment and rapidly dropped upon auxin withdrawal (Fig. 4E).”

May combine two claims.

The 64 cohesin-independent loop anchors often formed focal interchromosomal links between pairs and large cliques of interactions.

“We noticed that our 64 cohesin-independent loop anchors determined from our 61 loop intrachromosomal list often formed focal interchromosomal links between pairs of loop anchors and that there were large cliques of interactions between anchors (Fig. 4A,E, 5A).”

Probably not a substantial claim: may not pass S3 (load-bearing).

The treated higher-order contact data set contained 25M triples, 14M quadruples, and 55K quintuples.

“With this procedure, we obtained 32M triples, 18M quadruples, and 75K quintuples in our untreated data set, and 25M triples, 14M quadruples and 55K quintuples in our treated data set.”

Probably not a substantial claim: may not pass S3 (load-bearing).

A time-course APA at 100kb resolution using all intrachromosomal pairs of the 47 superenhancer-overlapping anchors in the 88-loop automated list gave similar results.

“Similar results were seen upon performing the time course APA at 100kb resolution using all intrachromosomal pairs of the 47 superenhancer overlapping anchors in the 88-loop automated list from above (Fig. S4E).”

Probably not a substantial claim: may not pass S3 (load-bearing).

The alternating construct tended to adopt an overall figure-8 shaped fold or hierarchical loop.

“The alternating construct tends to adopt an overall figure-8 shaped fold or hierarchical loop (Grigoryev et al., 2016).”

• cites Grigoryev et al., 2016 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

1.4% of the 2,145 significantly changed non-expressed genes were significantly downregulated with more than a 1.3-fold change.

“1.4% of these genes were significantly downregulated (>1.3-fold change), but it is unclear what reductions in expression at such low levels of expression mean biologically.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The authors observed high rates of quadruples but not many quintuples.

“While we did not observed many quintuples, we did observe high rates of quadruples, so we devised several strategies to project quadruples onto triple space in order to use them along with our triples in searching for hubs.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The untreated higher-order contact data set contained 32M triples, 18M quadruples, and 75K quintuples.

“With this procedure, we obtained 32M triples, 18M quadruples, and 75K quintuples in our untreated data set, and 25M triples, 14M quadruples and 55K quintuples in our treated data set.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Of 14,853 genes not expressed in untreated cells, 2,145 were significantly changed by DESeq2.

“To look for signs of ectopic activation, we examined the 14,853 genes that were not expressed (RPKM<0.5) in untreated cells. We identified 2,145 genes that were significantly (adjusted p<0.05) changed by DESeq2.”

May combine two claims.

In contact maps of the alternating constructs, the densest regions came from wild-type/wild-type local and nonlocal interactions, followed by local folded/folded and wild-type/folded interactions.

“The most dense regions in the contact maps of the alternating constructs come from wt/wt local and nonlocal interactions. These are followed by the local folded/folded and wt/folded interactions.”

The quote only partly states this claim.

They identified 203 interchromosomal cohesin-independent links enriched at least 5.5-fold over local background.

“Using this procedure, we identified 203 interchromosomal cohesin-independent links.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Among expressed genes, 2 genes showed a 5-fold change.

“Stronger effects were seen, but less frequently: 64 genes (0.5%) showed a 2-fold change, and 2 genes showed a 5-fold change (Fig. 7B).”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

In the control analysis of expressed genes, 12,191 genes were examined and 4,251 were significantly changed by DESeq2.

“We next looked for changes in the 12,191 genes that were expressed (RPKM>0.5) in untreated cells (Fig. 5B). We identified 4,251 genes that were significantly changed (adjusted p<0.05) changed by DESeq2.”

The quote only partly states this claim.

The distribution of distance to the nearest superenhancer for the control downregulated genes was shifted significantly closer compared to randomly selected genes.

“The overall distribution of distance to the nearest superenhancer was shifted significantly closer compared to randomly selected genes (Fig. S7D).”

Probably not a substantial claim: may not pass S3 (load-bearing). The quote only partly states this claim.

In the control, 86 genes (1%) showed a 2-fold change.

“Stronger effects were seen, but less frequently: 86 genes (1%) showed a 2-fold change, and 3 genes showed a 5-fold change (Fig. S7C).”

Probably not a substantial claim: may not pass S3 (load-bearing).

Of the control non-expressed genes, 1,179 (7%) exhibited leaky transcription in treated cells.

“In addition, 7% of these genes (1179) exhibited “leaky” transcription in treated cells: a larger PRO-Seq signal (p<0.05, >1.3 fold change difference) that fell short of the threshold for an expressed gene (i.e., RPKM was still below 0.5).”

Probably not a substantial claim: may not pass S3 (load-bearing).

In the control, 43 genes were 2-fold downregulated with p<0.05 after auxin treatment.

“We identified 43 genes that were 2-fold downregulated with p<0.05 after auxin treatment.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Of the control expressed genes, 1,861 (15%) showed a larger transcriptional effect with p<0.05 and more than 30% change in RPKM.

“The remaining genes (15%, 1861) showed a larger transcriptional effect (p<0.05, >30% change in RPKM).”

Probably not a substantial claim: may not pass S3 (load-bearing).

1.8% of the control non-expressed genes were strongly downregulated with more than a 1.3-fold change.

“1.8% of these genes were strongly downregulated (>1.3-fold change), but it is unclear what reductions in expression at such low levels of expression mean biologically.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Of the remaining three associated genes, two were downregulated.

“Of those 15 loops and the 12 genes associated with them, 9 were not expressed in HCT-116 (of the other three, two were downregulated, and one was upregulated).”

Probably not a substantial claim: may not pass S3 (load-bearing).

Of the 43 control downregulated genes, 28 were within 500kb of superenhancers.

“We noticed that many of the genes that were downregulated (by >2-fold) were located within 500kb of superenhancers (28 of 43).”

Probably not a substantial claim: may not pass S3 (load-bearing). The quote only partly states this claim.

In the control, 3 genes showed a 5-fold change.

“Stronger effects were seen, but less frequently: 86 genes (1%) showed a 2-fold change, and 3 genes showed a 5-fold change (Fig. S7C).”

Probably not a substantial claim: may not pass S3 (load-bearing).

Of the 43 control downregulated genes, 21 (49%) were within 500kb of one of the top 100 superenhancers.

“Of these genes, 49% (21 of 43) were located with 500kb of one of the top 100 superenhancers.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Of the remaining three associated genes, one was upregulated.

“Of those 15 loops and the 12 genes associated with them, 9 were not expressed in HCT-116 (of the other three, two were downregulated, and one was upregulated).”

Probably not a substantial claim: may not pass S3 (load-bearing).

The data are consistent with a previous report that genome compartmentalization is preserved after depletion of cohesin or CTCF.

“Our data is consistent with a previous report that genome compartmentalization is preserved after depletion of cohesin (Seitan et al., 2013) or CTCF (Nora et al., 2017).”

• cites Seitan et al., 2013 sourceNot checked yet

  Not checked in the free run.

• cites Nora et al., 2017 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

Deletion of CTCF sites interferes with loop and contact domain formation.

“Consistent with this, deletion of CTCF sites interferes with loop and contact domain formation (Guo et al., 2015; Sanborn et al., 2015; de Wit et al., 2015).”

• cites Guo et al., 2015 sourceNot checked yet

  Not checked in the free run.

• cites Sanborn et al., 2015 sourceNot checked yet

  Not checked in the free run.

• cites de Wit et al., 2015 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

Time-resolved maps were not sufficiently deep to assess the 102 loop domains shorter than 100 kb.

“(The time-resolved maps were not sufficiently deep to assess the 102 loop domains shorter than 100kb.)”

May combine two claims.

Rate of loop domain re-formation varies greatly across the genome and is associated with factors ranging from cohesin loading to local regulatory activity.

“Taken together, these results indicate that the rate of loop domain re-formation varies greatly across the genome, and is associated with factors ranging from cohesin loading to local regulatory activity.”

Classified as probably not a substantial claim.

CTCF and cohesin help regulate genome folding.

“This suggests that these proteins help regulate genome folding (Merkenschlager and Nora, 2016).”

• cites Merkenschlager and Nora, 2016Not checked yet

  Not checked in the free run.

Classified as probably not a substantial claim. May combine two claims.

CTCF and cohesin colocalize on chromatin and are located at loop anchors and contact domain boundaries.

“Many studies have shown that the insulator protein CTCF and the ring-shaped cohesin complex colocalize on chromatin (Wendt et al., 2008) and lie at the anchors of loops (Rao et al., 2014; Splinter et al., 2006) and the boundaries of contact domains”

• cites Wendt et al., 2008 sourceNot checked yet

  Not checked in the free run.

• cites Rao et al., 2014 sourceNot checked yet

  Not checked in the free run.

• cites Splinter et al., 2006 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

It is unclear what effect NIPBL depletion would be expected to have on loop domain formation because NIPBL may not be essential for cohesin loading.

“In any case, it is unclear what effect NIBPL depletion would be expected to have on loop domain formation because, although NIPBL facilitates cohesin loading, it may not be essential for cohesin loading.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

These findings do not rule out a central role for protein chaperones in vivo.

“Of course, these findings do not rule out a central role for protein chaperones in vivo.”

May combine two claims.

While cohesin plays a role in preventing ectopic activation, most genes remain inactivated even in the absence of cohesin and loop domains.

“Thus, while cohesin plays a role in preventing ectopic activation, most genes remain inactivated even in the absence of cohesin and loop domains.”

May combine two claims.

Cohesin-independent loops resemble superloops on the inactive X chromosome in being very large, not forming contact domains, having anchors that form cliques, and being marked by H3K27 acetylation.

“In many respects, the cohesin-independent loops resemble the superloops we previously observed on the inactive X chromosome (Darrow et al., 2016; Rao et al., 2014): they are very large (up to 77Mb), the intervals they span do not form contact domains, and their anchors tend to form cliques and are marked by H3K27 acetylation (Fig. S4F).”

• cites Darrow et al., 2016 sourceNot checked yet

  Not checked in the free run.

• cites Rao et al., 2014 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Deletion of WAPL results in formation of thousands of new loops and loop domains, which are larger than those found when WAPL is intact.

“Finally, an exciting recent experiment demonstrated that deletion of WAPL, a cohesin antagonist that removes cohesin from chromatin, results in the formation of thousands of new loops and loop domains, which are larger than those found when WAPL is intact (Haarhuis et al., 2017).”

• cites Haarhuis et al., 2017 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Cohesin is required for the establishment and maintenance of loop domains.

“We find that cohesin is required for the establishment and maintenance of loop domains.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Their data for cohesin depletion is consistent with the study by Nora et al., 2017.

“Our data for cohesin depletion is consistent with is study.”

• cites Nora et al., 2017 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

The discrepancy between early studies and their results may be due to low resolution Hi-C not distinguishing loop domains from compartment domains, and to incomplete cohesin depletion.

“The discrepancy may be due to (i) the fact that low resolution Hi-C analysis cannot not distinguish between loop domains, which are sensitive to cohesin depletion, and compartment domains, which are not; and (ii) the possibility that the cohesin depletion was incomplete (Fig. S2C).”

May combine two claims.

Cohesin-associated loops may both facilitate activation of promoters by their distal enhancer elements and block activation by inappropriate enhancers, with the former being somewhat more common.

“—suggesting that cohesin-associated loops may both facilitate activation of promoters by their distal enhancer elements and block activation by inappropriate enhancers, with the former being somewhat more common.”

May combine two claims.

After cohesin loss, compartment structure is strengthened in the absence of cohesin, as loop domains spanning multiple compartment intervals lead to mixing among loci in different compartments.

“(ii) compartment structure is strengthened in the absence of cohesin, as loop domains spanning multiple compartment intervals lead to mixing among loci in different compartments;”

Probably not a substantial claim: may not pass S3 (load-bearing).

Cohesin loading independent of NIPBL has been observed in vitro, albeit at low efficiency.

“Indeed, cohesin loading independent of NIPBL has been observed in vitro, albeit at low efficiency (Davidson et al., 2016; Stigler et al., 2016).”

• cites Davidson et al., 2016 sourceNot checked yet

  Not checked in the free run.

• cites Stigler et al., 2016 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

The data are consistent with the observation that genome compartmentalization is weakened in cells where WAPL is knocked out, leading to larger loop domains.

“Our data is consistent with the observation that genome compartmentalization is weakened in cells where the cohesin-antagonist WAPL is knocked out, leading to larger loop domains (Haarhuis et al., 2017).”

• cites Haarhuis et al., 2017 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

RNA polymerase II can push cohesin along DNA, but elongation rate estimates (9-90 bp/s) are slower than what we observe.

“Similarly, RNA polymerase II can push cohesin along DNA (Davidson et al., 2016) but elongation rate estimates (9–90 bp/s; (Jonkers and Lis, 2015)) are slower than what we observe.”

• cites Davidson et al., 2016 sourceNot checked yet

  Not checked in the free run.

• cites Jonkers and Lis, 2015Not checked yet

  Not checked in the free run.

Classified as probably not a substantial claim.

To rule out mitotic stalling, the authors repeated Hi-C after synchronizing and arresting cells at G1/S.

“In order to ensure that our results were not due to the HCT-116 cells stalling in mitosis, we also repeated our Hi-C experiments after cell synchronization and arrest of the cells at the G1/S boundary.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Hi-C libraries from synchronized, arrested cells gave comparable results.

“Similar results were obtained with Hi-C libraries from synchronized and arrested cells (Data. S1, I)”

May combine two claims.

Loop domains disappear shortly after auxin-induced cohesin loss and reappear shortly after auxin withdrawal, implying they require cohesin both for formation and maintenance and do not represent stable states of chromosome condensation.

“We show that loop domains disappear shortly after auxin-induced cohesin loss, and reappear shortly after auxin withdrawal, implying that (i) they require cohesin both for formation and maintenance; and (ii) they do not represent stable states of chromosome condensation.”

May combine two claims.

The data are less consistent with superenhancer loops forming by extrusion and more consistent with an alternative mechanism, possibly facilitated diffusion or phase separation.

“Our data is therefore less consistent with a model where where superenhancer loops form by extrusion, and more consistent with the presence of an alternative mechanism, perhaps based on some form of facilitated diffusion or phase separation (Sanborn et al., 2015).”

• cites Sanborn et al., 2015 sourceNot checked yet

  Not checked in the free run.

• AssumptionRests on an assumption: Loop extrusion cannot create links between loci on different chromosomes.

Probably not a substantial claim: may not pass S1 (asserted), S3 (load-bearing).

The underlying physical process is hypothesized to be loop extrusion.

“We (Sanborn et al., 2015) and others (Alipour and Marko, 2012; Fudenberg et al., 2016; Nasmyth, 2001) have hypothesized that the underlying physical process is the formation of loops by extrusion.”

• cites Sanborn et al., 2015 sourceNot checked yet

  Not checked in the free run.

• cites Alipour and Marko, 2012Not checked yet

  Not checked in the free run.

• cites Fudenberg et al., 2016 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

Their results are consistent with these findings.

“Our results are consistent with these findings.”

Probably not a substantial claim: may not pass S3 (load-bearing).

To ensure high-quality annotation, H3K4me1 and H3K27Ac peaks required adjusted -log10 p-values of at least 7 and 5.

“Finally, for H3K4me1 and H3K27Ac, we filtered peaks called by MACS by requiring an adjusted -log10 p-value of at least 7 and 5 respectively in order to ensure that we had a high quality peak annotation.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Every ChIP-Seq experiment was run alongside whole-cell extract input controls.

“All ChIP-Seq experiments were processed in parallel with whole cell extract input controls.”

Probably not a substantial claim: may not pass S3 (load-bearing).

These estimates are similar to ones obtained studying the SMC complex in B. subtilis [500-1000 bp/s].

“These estimates are similar to ones obtained studying the SMC complex in B. subtilis [500–1000 bp/s] (Wang et al., 2017).”

• cites Wang et al., 2017 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Cohesin-associated looping raises contacts inside loop domains and disrupts compartmentalization, allowing otherwise separated elements to mix.

“Our study suggests a model where cohesin-associated looping, by increasing the frequency of contact between loci within loop domains and by disturbing patterns of compartmentalization, facilitates mixing between elements (such as genes and superenhancers) that would otherwise be segregated.”

May combine two claims.

This loop is extruded at an average rate of no less than 375 bp/s, with each Xcom subunit sliding at no less than 188 bp/s.

“Thus, this loop is extruded at an average rate of no less than 375 bp/s, with each Xcom subunit sliding at no less than 188 bp/s.”

Classified as probably not a substantial claim.

Prior work with similar methods found a strong link between having a loop domain and large expression increases for genes at its anchor.

“earlier studies, using similar methods, that showed a strong correlation between the presence of a loop domain and many-fold increases in the expression of genes at the loop domain anchor (Kagey et al., 2010; Rao et al., 2014).”

• cites Kagey et al., 2010 sourceNot checked yet

  Not checked in the free run.

• cites Rao et al., 2014 sourceNot checked yet

  Not checked in the free run.

Classified as probably not a substantial claim.

In the loop extrusion model, loop domains form when a cohesin-based extrusion complex binds chromatin and its subunits slide in opposite directions until they reach a bound CTCF protein.

“In this model, loop domains form when a hypothetical cohesin-based extrusion complex (“Xcom”), which comprises two physically tethered subunits, binds chromatin at a particular location; subsequently, the subunits slide in opposite directions until they arrive at a bound CTCF protein.”

• cites Sanborn et al., 2015 sourceNot checked yet

  Not checked in the free run.

Classified as probably not a substantial claim.

PRO-Seq in parental CMV-OsTIR1 cells with or without auxin served as a control for auxin and RAD21-tagging effects on transcription.

“By performing PRO-Seq on CMV-OsTIR1 cells with and without auxin treatment, we could control for transcriptional effects of the auxin treatment itself on HCT-116 cells, as well as any consequences of tagging the RAD21 protein.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Small rises in receptor tyrosine kinase gene expression can meaningfully affect cell proliferation.

“For example, modest increases in the expression of receptor tyrosine kinase genes can have meaningful effects on cell proliferation (Flavahan et al., 2016).”

• cites Flavahan et al., 2016 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

Given the correlation between loop reformation speed and Nipbl, loading time may be a significant consideration.

“(Given the correlation between loop reformation speed and Nipbl, loading time may be a significant consideration.)”

May combine two claims.

Disappearance of cohesin can eliminate all loop domains without influencing CTCF binding.

“Thus, the disappearance of cohesin can eliminate all loop domains without influencing CTCF binding.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The data are consistent with several models of cohesin-dependent loop domain formation.

“The data presented above are consistent with several models of this process.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

WAPL deletion seems to raise Xcom processivity, enlarging loops.

“Similarly, deletion of WAPL appears to increase the processivity of the Xcom, and thus increases the size of loops (Haarhuis et al., 2017).”