alexxijielu
commented
2 years ago Hi Jackson,
Thanks, I'm glad you like the work!
We applied these filters at different times:
- The raw data should have the homologues filtered by redundancy. I believe this is done at the protein-level, not the IDR level (CCing my collaborator Iva to double-check this), so you might still get some IDRs with all gaps depending on the alignment.
- After, we filter IDR sequences (for "X" characters and under 5 AAs) in the data loader itself.
Hope this helps.
Alex
From: jacksonh1 @.> Sent: Friday, November 11, 2022 5:20 PM To: alexxijielu/reverse_homology @.> Cc: Subscribed @.***> Subject: [alexxijielu/reverse_homology] Question about human homologous IDR dataset (Issue #2)
Hello Alex,
I quite like your paper, and the work that you've done is very exciting! I think that your dataset of homologous IDRs in the human proteome could be a very valuable resource for me in some of the research I'm doing. I've been looking at the most recent dataset deposited to zenodo (V6 - 6311384https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fzenodo.org%2Frecord%2F6311384&data=05%7C01%7Clualex%40microsoft.com%7Cd785a2a7af2c46c5783008dac432d679%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C638038019874091879%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=COYq1xY4N4R754sQXHMsJinQT1gJ8Ga6EmJQVLOPLjg%3D&reserved=0) (file - human_idr_homologues.zip) and had just a few questions about it. Are these the homolog IDRs after applying the evolutionary distance-based clustering/filtering method described in the methods to remove "redundant" sequences? The methods also state that sequences with "X" characters or that were 3x longer/shorter than the human protein were removed. However, I've found cases in the dataset where there are sequences with 'X' characters or with empty sequences present (See example below). It is easy for me to filter those out on my end, but it did make me wonder if the clustering/redundancy filter has been applied to these sequences or not?
example from HUMAN13574_301to347.fasta:
ACAM105111
-Q-------G--------------------------------------------------
CHAP605088
-K----------------------------------------------------------
SYNY302936
-K-------K--------------------------------------------------
PROMS00447
PROM400393
-C----------------------------------------------------------
PROM200471
PROM000417
PROM302117
PROM900417
PROMM00236
PROM500458
PROM100444
-E-------EVST-----------------------------------------------
PROMT00426
-E-------EV-------------------------------------------------
PROMA00393
PROMP00444
CYAGP02842
SYNE702106
SYNP602035
SYNP302362
-Q-------IAND-----------------------------------------------
Thanks so much for the help, Jackson
- Reply to this email directly, view it on GitHubhttps://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Falexxijielu%2Freverse_homology%2Fissues%2F2&data=05%7C01%7Clualex%40microsoft.com%7Cd785a2a7af2c46c5783008dac432d679%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C638038019874091879%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=5gV3D8lX4Y0ORWNg0gyAYbmlgnZW9SGtXK3rDcmd2UY%3D&reserved=0, or unsubscribehttps://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAE2KQENSB4OPDDYXMPSCMODWH3BABANCNFSM6AAAAAAR57KM4I&data=05%7C01%7Clualex%40microsoft.com%7Cd785a2a7af2c46c5783008dac432d679%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C638038019874091879%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=JKLK8tLvtdf4LTHnAqRrb25oxxxGZZfT7N7cpv46MKM%3D&reserved=0. You are receiving this because you are subscribed to this thread.Message ID: @.**@.>>
jacksonh1
Hello Alex,
I quite like your paper, and the work that you've done is very exciting! I think that your dataset of homologous IDRs in the human proteome could be a very valuable resource for me in some of the research I'm doing. I've been looking at the most recent dataset deposited to zenodo (V6 - 6311384) (file - human_idr_homologues.zip) and had just a few questions about it. Are these the homolog IDRs after applying the evolutionary distance-based clustering/filtering method described in the methods to remove "redundant" sequences? The methods also state that sequences with "X" characters or that were 3x longer/shorter than the human protein were removed. However, I've found cases in the dataset where there are sequences with 'X' characters or with empty sequences present (See example below). It is easy for me to filter those out on my end, but it did make me wonder if the clustering/redundancy filter has been applied to these sequences or not?
example from
HUMAN13574_301to347.fasta:Thanks so much for the help, Jackson