DIAMOND and BLAST version issue #41
Comments
|
Hello @Yliub This error message means that when HGTector calls DIAMOND to do a homology search, the latter program failed for some reason. The error code "1" does not tell us much about the actual reason though. I would suggest that you try the following steps:
diamond blastp --query input.fa --db /path/to/db --outfmt 6 qseqid sseqid pident evalue bitscore qcovhsp staxids |
|
Hi Qiyun, |
|
If the CPU and memory have a effect on this phenomenon? my computer memory is 8Gb and CPU is dual core 4 thread |
|
Hello @Yliub It may be the case that 8Gb memory is not sufficient for DIAMOND. What do you see if you run the command manually as I suggested above? And what do you mean by "conduct hgrector search by remote server"? I guess the wrong message won't be the same. Can you let me know it? |
|
Hi, |
|
@Yliub Sorry. No. This database may require more than 8Gb. Can you try that command? |
|
ok，I will reply you as soon as I come to my work room because my side is two o'clock now，thank you so much. |
|
You are welcome! :) |
|
Hello @qiyunzhu is it necessary that the computer memory must be more than 8Gb if I run remote search? |
|
Hi@qiyunzhu |
|
Hi @Yliub Thanks for the error message! I believe this is a bug, and I just fixed it. Can you reinstall HGTector2 and test if the problem is resolved? Thank you! |
|
Hi @qiyunzhu I reinstall the hgtector2 pepiline and run hgtector remote search command. the screen back me the following: |
|
Hi @qiyunzhu before the error messagae, the pepline has ran some time, and retrieved a part of the result. is that as a result of the internet speed limit? |
|
Hello @Yliub Sorry for the late response! It seems to me that the pipeline worked; and the error was due to Internet limit. If you run the same command again, the pipeline will resume from the checkpoint. Hopefully, you will get the final result after one or a few trials :) But the |
|
Hi @qiyunzhu, |
|
Hello @Yliub , I think there may be a solution after carefulling reading the DIAMOND manual. It says:
So, you may follow this instruction and the HGTector database compilation instruction to build a new DIAMOND database which demands less than 8Gb memory on your computer. By the way, I just fixed a bug of HGTector2. You may update your copy by running: |
|
@qiyunzhu Ok, thanks for your help. |
|
Hi@qiyunzhu, I try to execute the --block-size/-b command in diamond algorithm. but it doesn‘t work, I carefulling reading the DIAMOND manual, I think the --block-size/-b command should be executed during making alignment process. so could you please make it inbuilted in your hgtector search command? |
|
Hello @Yliub Thanks for letting me know! Running DIAMOND on a computer with 8GB memory is something unexpected and could reduce performance. So I hesitate to make the change you suggested into HGTector. That being said, you can modify your HGTector installation to include this feature. Specifically, you can edit this line in search.py to add this DIAMOND parameter. Hope it works! |
|
@qiyunzhu ok，I will try it， thank you very much. |
|
Dear@qiyunzhu, I am sorry for disturbing you again, I modified the search.py file in the line you mentioned and subsequently run hgtector search command, but it seem doesn't work, and the error meassage is showed in the attached picture. if possible, could you please add the --block-size parameter in the corresponding search.py line, it's appreciate that if you can help me to add this parameter and send the revised file to me. my e-mail is [email protected]. |
|
Hi@qiyunzhu, by the way, I run the command " diamond blastp --query input.fa --db /path/to/db --outfmt 6 qseqid sseqid pident evalue bitscore qcovhsp staxids" plus " --block-size" aligner options, it works successfully agianst the local database. so I think the "hgtector search" using diamond with " --block-size" aligner options will work. |
|
Hello @Yliub , in the way you modified the code, you added an attribute cmd = '{} blastp --block-size 1.0 --query {} --db {} --threads {} --tmpdir {}'.format(I have sent a modified |
|
@qiyunzhu hi，it works, thanks for your kind help |
|
Hi @qiyunzhu, I am sorry for disturbing you again. the HGTector, as you known, is a powful tool for the predication of HGT events, my 30 bacterial genomes were used as a whole to predict the gene donors using default parameters. but the results I think were untrusted (I attached the results in the figure uploaded). the predicted gene donors were somtimes phylogenetic included in the upper level, for example Serratia sp. S1B belongs to Gammaproteobacteria, and the Gammaproteobacteria were also predicted, and all the predicted gene donors can eventually belongs to bacteria. so my puzzle is that how can I predicted the gene donors at a uniform bacterial level (such as Phylum，Class，Order or Family). thank you very much. Best regards to you |
|
Hello @Yliub Sorry for the late response. The beginning of 2020 has been quite tough in my calendar. Re. your question: The scenario you described is not unusual. HGTector does not explicitly predict donors, but only reports a taxonomic unit that could summarize the top hits from the distal group. The details are here. This algorithm is consistent with DIAMOND. Now here comes the question: if these "top hits" are too diverse in taxonomy, HGTector will just find the LCA, which can be at a very high level, and thus less meaningful. However this is the best the program can do given current search results. It is safe, though not precise. Please don't expect that every gene will have a donor at a lower level (e.g., species). If the program says "Gammaproteobacteria", it means that although Serratia sp. S1B could be a plausible donor, other gammaproteobacteria also have high bit scores, and the program cannot safely exclude them from the candidate donor list, therefore it chooses to report the LCA. Please note that none of the taxonomic names shown here are the true donor, they are but close relatives to the true donor, which must be an organism once lived in history and died out already. The goal of donor prediction is to describe the hypothetical donor using extant taxonomy. If you really want to get precise donors (e.g., genus- or species-level), you can add If you want to force all donors at a specific rank, it is also doable: by translating the donors into a specific rank. Actually the previous HGTector can do it, but I dropped this functional module in HGTector 2. Maybe I should add it back. Hope this helps! |
|
@qiyunzhu So, first of all, thank you for your patient and professional guide. I don't understand what you have occured, but wishing you have a good day so as to pack your mood and work forward. Now, let's back to the questions. I just want to care about the hypothetical donors at a specific rank, though, you known, they are not true donor, so could you please add corresponding function module in the latest updated HGTector 2. Thank you very much. Best regards to you |
|
Hello @Yliub thanks! Same wishes to you. I am totally fine. Just busy working. I think this function is useful and I will add in the next version. It may not be too soon since I have got other priorities. Thanks for your interest. |
|
@qiyunzhu OK, that's fine, please inform me when the new version is available. thanks! |
|
Hi @ @qiyunzhu, have you add the functional module which can predicte the hypothetical gene donors at at a specific rank in the newest HGTector version ？ |
Hi Qiyun,
I am using the HGTector2 pipeline to investigate the HGT event of my 30 bacterial strain. I set up my local diamond database using the file downloaded from your dropbox. I seem that everything is ok. but when I use the search function, the screen always traceback me a messages showed as follow:
Traceback (most recent call last):
File "/home/liub/miniconda3/envs/hgtector/bin/hgtector", line 100, in
main()
File "/home/liub/miniconda3/envs/hgtector/bin/hgtector", line 39, in main
module(args)
File "/home/liub/miniconda3/envs/hgtector/lib/python3.7/site-packages/hgtector/search.py", line 167, in call
else None)
File "/home/liub/miniconda3/envs/hgtector/lib/python3.7/site-packages/hgtector/search.py", line 706, in search_wf
res = self.diamond_search(seqs)
File "/home/liub/miniconda3/envs/hgtector/lib/python3.7/site-packages/hgtector/search.py", line 1542, in diamond_search
raise ValueError('diamond failed with error code {}.'.format(ec))
ValueError: diamond failed with error code 1.
what's wrong with this? waiting for your repy. thank you very much
Regards
The text was updated successfully, but these errors were encountered: