On Monday and Tuesday we will release final domains from the RNA polymerase that we started predicting this week.

On Thursday and Friday we will have two heteromeric targets (one CAPRI) together with their subunits.

On Wednesday we will finish the polymerase by suggesting the whole structure for prediction (~2,200 res). Even though this is a monomeric target, we will be assessing models similarly to those submitted on oligomeric targets. In other words, we will not be assessing accuracy of prediction on individual domains (as these will be evaluated separately), but rather assess ability of predictors to correctly arrange the domains in a complete structure. To distinguish this target from other monomeric targets, we will assign it a target name starting from H (like for hetero-meric targets). We will be sending requests for automatic modeling of this target only to servers registered for multimeric prediction. Contact predictions will not be assessed for this target, but rather for individual domain targets (i.e., T1031, 33...). If you have a multimeric server that also does RR prediction, please make sure that it does not send us an RR prediction for this target, especially the whole distance map as this will have taken around 2,500,000 lines of text, which translates into files around 200 MB in size - no email client will handle files this big. In general, it is a good idea to limit your contact predictions to 50,000 lines with only the most important and reliably predicted contacts included.

---

Week 4 plans.

On Monday we will release a heteromeric complex. We know that the complex has 1:1 component ratio, but a higher copy number N, which we do not know. So, besides predicting a hetero-dimer, another interesting thing would be to predict the whole complex as a homo-N-mer of hetero-dimers.

Tuesday and Wednesday will see two targets each; Thursday and Friday only one.

CAPRI participants will get two targets to predict on Wednesday and Friday.

In consultations with the EMA community and the refinement assessor, we are in the process of selecting targets for refinement from those modeled in CASP-covid. Once we come to an agreement (tentatively early next week), those will be released for prediction. Note that these targets are open for prediction not only to our refinement registrants, but also to all CASP14 predictors. The principles of how these targets will be assessed will be worked out by the assessors.

---

Week 5 plans.

Monday through Wednesday we will release two targets each day, including one for CAPRI. On Thursday and Friday we will have sub-complexes of a large heterocomplex, including at least one more CAPRI target.

Last week our predictors saw first three refinement targets selected from those modeled in the CASP-covid experiment (names starting from R19). We are reminding that that these targets are open for prediction not only to our refinement registrants, but also to all CASP14 predictors. We will see two more refinement targets of this type on Monday and Tuesday and then we will start releasing refinement targets selected from among those released earlier in the regular CASP14 experiment.

Last week three large targets (T1050, T1052 and T1053) were released on three consecutive days causing increased load on automatic servers. In order to allot the servers extra time to process these large targets, we are extending server prediction deadline for two of the three targets (T1050 and T1051) for one extra day.

---

Week 6 plans.

1. We have moved the deadline for targets H1046, T1046s1 and T1046s2 to June 23, which is three days earlier than originally scheduled. The structure of this complex will be publicly released by the PDB next week.

2. Next week we will have:
Monday - 2 targets (including one CAPRI)
Tuesday - 1 heterocomplex and its subunits
Wednesday - 1 heterocomplex and its subunits (CAPRI)
Thursday - 2 targets
Friday - 1 heterocomplex and its subunits (CAPRI)

3. We will continue releasing refinement targets at the rate of one target per day.

In the previous CASP competitions, R@H inserted the name of target (es T1046) in the name of wus.
Seems that now this is not

---

week 7 (June 29 - July 5) (2020-6-28)
Notes from the previous week

1. Target H1046 cancellation.
Last week we learned that target H1046 was scheduled for release by the PDB three days ahead of its planned CASP prediction deadline. We tried to save the target by moving its deadline prior to the PDB release date (June 23). However, this did not fully help as we learned that a paper describing the complex was published online a few days earlier (June 17, https://www.sciencedirect.com/science/article/pii/S0022283620304095). Since the vast majority of H1046 predictions were received after June 17, we are forced to cancel this target. Even so, we do not plan to immediately cancel the subunits of this target as around 2/3 of all the received predictions were submitted before June 17. Assessors will carefully inspect the predictions and their submission dates and will make a final decision in due course.

2. Refinement targets with multiple starting models.
Last Friday we released two refinement targets (R1040v1 and R1040v2) that correspond to the same regular target, T1040, but different starting models. In previous CASPs we learned that some targets are more refine-able than others, even though reasons for that were not always obvious. By taking out of the equation the target refine-ability factor, we hope to better understand the extent to which the final refinement result depends on the accuracy of the starting model. Note that for such double-starting targets we will not be revealing accuracy of starting models or location of the areas needing most attention. We plan to release several refinement targets of this kind in CASP14, with the next one coming out already this Monday.


Plans for the next week

Monday - 1 target (CAPRI)
Tuesday - 1 target
Wednesday - 1 heterocomplex and its subunits
Thursday - 2 targets, including one CAPRI
Friday - 1 target

We will continue releasing refinement targets at the rate of one target per day. If double-starting targets will be released in one day, their expiration dates will be spread in time, whenever possible.

---

Week 9 (July 13 - July 19) Notes on the previous week targets.

1. T1080 - only 133 residues.
We learned that the structure of this target contains only 133 residues out of the 922 posted. So, the predictors can save their resources by concentrating on modeling residues 649-781. If you are doing multimeric modeling - all chains of the multimer are restricted to the specified residues.

2. H1081 - assembly only.
This target is a homo-multimeric A20 structure that will be evaluated only in the assembly context. Related structure of a decamer exists, so the challenge here is to predict the interface between the stacked decamers and to model the precise fold.

Plans for the next week - 8 targets:

Monday - 2 targets, including one CAPRI
Tuesday/Wednesday - 1 target
Thursday - 2 targets, including one CAPRI
Friday - 2 targets

---

CASP14 - week 10 (July 20 - July 26) and week 11 (July 27 - Aug 2) (2020-7-18)
1. Tertiary structure /Assembly /Contacts
We are approaching the end of the regular CASP14 season. Only two full weeks of prediction are ahead of us. As of today, we have 13 targets left, with one or two more still probable. We can potentially release them in two weeks, but since the majority of the remaining targets are large, we prefer to release only one target per day in the coming two weeks. The remaining targets (3 as it stands now) will be released in the first week of August. There will be no CAPRI targets next week, and one target (probable) the week after that.

2. There will be 5 refinement targets next week.

3. As you might have noticed, we released our first data-assisted target on Friday, S1063. The SAXS data for this target came from the author of the structure, so they may be in slightly different format than those released in CASP13, even though we tried to prepare the data in the form close to that of CASP13. This target is interesting in the way that it has not been solved yet, even though the crystallographer working on it has collected good experimental data. We hope that CASP models (especially data-assisted for the multimeric complex) can help solving this structure. As it stands now, we do not have more data-assisted targets for you this month, however it is possible that we may obtain a few more next month.

---

CASP14 - week 11 (July 27 - Aug 2)
1. We are entering the last full week of CASP14 regular target release. This week we will have one regular target per day Monday through Thursday and two targets on Friday. This will leave us with only two more targets (as it stands today) for the overflow week of August 3. There will be more refinement targets next week.

2. Last week some predictors experienced issues with model submission on the large multi-chain assembly target H1060. To allow those CAPRI/CASP predictors to resubmit their models, we are extending the H1060 deadline until next Monday, July 27.

3. To allow sever groups without a large amount of computing resources finishing large last-week targets, we are extending by one day server deadlines for T1093 and T1095.

---

CASP14 - end of the regular target release (2020-8-4)
1. Oligomeric /tertiary structure /contact prediction.
Today we posted the last target for oligomeric prediction in CASP14. This concludes the CASP14 regular target release. Despite the coronavirus-related disruptions in the experimental lab operations, we were able to secure a diverse set of structures (including many FM or FM/TBM targets) for testing your prediction methods. All in all, 83 targets were offered for tertiary structure /contact prediction and 30 targets for oligomeric prediction (including 17 CASP/CAPRI targets).

2. Refinement.
We will continue releasing refinement targets throughout the next three weeks, however we do not expect more than five new targets in this time span. Besides that, we plan to reopen some of the closed refinement targets for further refinement to learn if longer MD runs can result in better CASP14 models.

3. H1103 (released today)
The last CASP14 target is a probable SARS-CoV2/Human complex. The components of this target are known (Orf3a, PDB:6xdc and HMOX1, PDB:1n3u), thus this is a pure assembly target where no tertiary structure /contact prediction are requested. An overview of this target can be found in our data repository: https://predictioncenter.org/download_area/CASP14/extra_experiments/

4. H1102 (released yesterday)
We have canceled this oligomeric-only target due to the diminishing hope for its structure determination.

---

CASP14 - week 13 (August 10-16) (2020-8-10)
1. Assembly target H1097, and its tertiary structure prediction subunits T1092-1096.

It was brought to our attention that a Ph.D. thesis describing some aspects of this structure is available online.
To level playing field, we are providing a link to this Ph.D. thesis: (https://www.skoltech.ru/app/data/uploads/2018/05/sokolovamaria_3414_62556_Maria-Sokolova_thesis_2018-4-06-181.pdf) and encourage predictors to check it.

We contacted the provider of H1097 and learned that most of the images in the thesis are models of known RNAP structures solved by other people. The thesis discusses only some fragments of the target structure that were available two years ago. The experimental data for the most interesting subunits of the complex were collected only in spring of 2020, and at the time of this writing the structure is in the final phases of model building, after which it will be refined. We are hopeful that the structure will be ready in time for the CASP14 assessment.

2. Refinement

Two new refinement targets will be released on Monday. Tuesday through Friday we will be re-releasing some of the already closed refinement targets to learn if longer refinement runs lead to more accurate models. These targets will be released with the same starting models as in the first round of the refinement, and will be marked with the 'x1' suffix in their name ('x' for extended).

3. EMA (QA)

This Tuesday we are finishing collecting estimates of model accuracy for regular CASP14 targets. However this will not end the EMA prediction season. After all QA predictions for regular targets are collected, we will release five more EMA targets as announced in the week 8 communique. These additional QA targets will be for SARS-CoV2 refinement targets released at the beginning of the CASP14 refinement season (names starting from R19).

---

CASP14 - week 15 (August 24-30); two NMR-assisted targets (2020-8-25)
1. TS /RR
Today (August 25) we closed the last CASP14 target in the tertiary structure /contact prediction categories. All in all, we collected around 48,000 predictions in these two categories.

2. Assembly
The last assembly target, H1103, will be closed this Saturday. Around 1,500 predictions have been submitted in this category to date.

3. QA
Accuracy estimation predictions were wrapped up on August 16 with a total of ~11,000 predictions.

4. Refinement
All CASP14 refinement targets have been released for prediction. We will continue accepting predictions on open targets until September 11.

5. NMR-assisted targets
Today we released the first out of two NMR-assisted targets (N1088). The data are provided at http://predictioncenter.org/download_area/CASP14/extra_experiments/. The second target will be released by the end of the week, tentatively on Thursday. This will end up all target release in CASP14.

6. We will start collecting Methods Abstracts next week and will be collecting them until September 15. Please start preparing abstracts already now so you will have enough time to meet the deadline.

---

CASP14 - week 16 (Aug 31- Sep 6) - Collecting methods abstracts. (2020-9-1)
1. Today we released the second of the two promised NMR-assisted targets, N1077. The data are provided at http://predictioncenter.org/download_area/CASP14/extra_experiments/. This concludes all target release in CASP14.

2. Now, when majority of CASP14 activity is over and the assessors started their work, it is time to submit write-ups of your CASP14 methods. We have opened a web page for collecting the extended abstracts. The submission deadline is September 15, 2020. Please read the instructions carefully (in particular those pertaining to formatting the first line of the Abstract) and address all the questions specified in the Abstract submission form (https://predictioncenter.org/casp14/abstracts.cgi).

3. As we announced earlier, CASP14 predictors' meeting will be online. Format of the meeting and dates of the sessions will be announced later this month. A program of the meeting will be available only after the assessors complete their analysis of models (tentatively, in November 2020).

---

Dear CASP community member:

As you no doubt realize, we will have to hold the CASP14 conference virtually. We are planning to do that from November 30th through December 4th. There will be a limited number of talks each day, lasting a total of not much more than 3 hours. All talks will be pre-recorded, and each session will be repeated to accommodate different times zones. Suggested session times are 16:00-19:30 CET for Europe and the US and 10:00- 13:30 JST for Asia. There will be separate local questions and discussion. We plan to make ‘posters’ and asynchronous interactions as effective as possible.

To compensate for the shorter and likely less intense meeting, we will encourage ongoing interest groups activities afterwards, and there will be a follow-up in-person meeting next summer, CoViD permitting.

We are still deciding the conference details, and would welcome suggestions on how to organize specific aspects of it, based on your own experience of virtual meetings and your preferences. Please include experience with handling a wide range of time zones, and tell us about any useful software and poster arrangements you have encountered.

---

Dear CASPers,

Starting last week and finishing this Monday, the CASP organizers had a series of online meetings with the assessors to hear the results of their analysis of CASP14 submissions. Some exciting results were reported and we are anticipating a very interesting conference! The assessors also suggested groups who should give talks at the conference. We expect to begin notifying selected speakers by the end of the week. The assessors have now been given the list of group identities (also available on the CASP14 web site) and may follow up with questions to individual participants.

We are currently working on the conference program and hope to have a working version by the end of next week. Registration for the conference should open by the end of this week. We will be requesting that all participants pay a nominal registration fee of $50 USD to cover expenses for the online tools used for conducting the conference. Stay tuned.

---

Impressive results from AlphaFold
CASP press release
DeepMind blog post