The Name on the Form
Mariana Quispe had been in the cleanroom since six, and the pink colonies on the plate were the wrong shape.
She had expected the usual halophile look: flat, angular, the color of a sunburn. These were rounder. Pinker. They looked like someone had spilled nail polish on the agar. She photographed them, labeled the plate, and went to the sequencing queue.
The queue had been running since Monday, and Mariana’s sample was still waiting. The first 16S run had come back Tuesday night, and she had read it at midnight in her apartment with a cup of tea gone cold.
The result was Halobacterium salinarum. Ninety-nine point one percent identity. The organism on her plate was not new. It was the thing that lived in the lab’s own water system, the one that showed up in the negative controls every quarter, the one the quality officer had taped a notice about to the tap.
She read the result three times. Then she opened the PI’s submission file, where Halobacterium donantis sat in a draft manuscript with a donor’s name in the acknowledgments and a Latinization note in the methods. The donor’s check had cleared in March. The registry deadline was Friday.
Mariana set the mug down and went back to the lab.
The PI was Dr. Serra. He was in his office, which was on the other side of the building, and he kept the door open because he liked to hear the corridor. Mariana stood in the doorway and told him the 16S result.
“Identity?”
“Ninety-nine point one. Against H. salinarum.”
He nodded. “V region?”
“V3 to V4.”
“The V regions are short,” he said. “You know that. They’re the barcode, not the organism. A contaminant can match a barcode and still be a species, if the rest of the genome carries the difference. That’s why we do the tree.”
Mariana had expected this. She had also expected the next part, which was that the tree had been built with a model trained on extremophiles, and extremophiles did not cluster like mesophiles, and the model knew a new species when it saw one.
“The model was trained on extremophiles,” Serra said.
“I know. I built it.”
“Then you know it knows.”
She did know. She had written the training script. The model weighted salt-tolerance genes heavily, because in extremophiles those genes were the signal. It was a reasonable choice. It was also the choice that made every pink halophile in the Atacama look like a new species, because every pink halophile in the Atacama shared salt-tolerance genes with every other pink halophile in the Atacama, and the model read that sharing as divergence.
She did not say this. She said, “I’d like to request the raw reads.”
Serra looked at her. “The reads are in the submission. I attached them.”
“The raw reads. The untrimmed ones. The ones before the pipeline.”
He leaned back. “Why?”
“Because the pipeline trims the barcodes. If the contaminant’s barcode is in the reads, the trimmed output will show a clean match. The untrimmed output will show where it came from.”
“And where will it have come from?”
“The water system.”
He smiled. It was the smile of a man who had heard this before. “Mariana. The water system is a source, not an explanation. Organisms travel. The Atacama is full of halophiles. The fact that one of them resembles a lab contaminant is not a reason to retract a name. It’s a reason to do more science.”
“I’m asking for the raw reads.”
“You have them.”
“I have the trimmed set. I’m asking for the raw set.”
He opened a drawer, took out a USB drive, and held it out. “These are the raw reads from the sequencer, before any processing. They are the same reads I used for the tree. If you find a water-system barcode in them, I will be the first to want to know.”
She took the drive. “Thank you.”
“Mariana.”
“Yes.”
“The name is good. The tree is sound. The donor’s money paid for the sequencing. If you find something, bring it to me. Don’t bring it to the registry.”
She nodded and left.
The raw reads took two days. She ran them through a different pipeline, one that did not trim, one that kept the adapter sequences and the barcode regions and the quality scores for every base. The reads were long. The sequencer had been a good one, and Serra had paid for long reads, which was the right choice for a genome announcement.
She found the barcode at position 14,203 in read 4,412. It was a 16S V3-V4 region, and it matched Halobacterium salinarum at ninety-nine point four percent. The match was in the untrimmed data, which meant it had been in the sample before any processing. It was not an artifact of trimming. It was in the organism.
The organism on her plate had the contaminant’s barcode in its genome. That could mean two things. It could mean the organism was the contaminant. Or it could mean the organism had acquired the contaminant’s barcode through horizontal gene transfer, which was common in halophiles, and the tree was right to separate them.
She wrote both possibilities in her notebook. Then she went to Serra.
“The barcode is in the raw reads,” she said. “Position 14,203. Ninety-nine point four.”
He looked at the printout. “Horizontal transfer.”
“Possibly.”
“Probably. The model accounts for it. The tree still holds.”
“The tree was built with the trimmed reads.”
“The tree was built with the same reads, trimmed. Trimming removes adapter sequence. It does not remove biological signal. Mariana, you know this.”
“I know it. I’m asking whether the model’s training on extremophiles might be over-weighting the salt-tolerance cluster and under-weighting the 16S divergence.”
He set the printout down. “The model has a validation accuracy of ninety-four percent on held-out extremophile genomes. I published that. You reviewed that. If you want to challenge the model, you write a paper. You do not challenge a name.”
She went back to her desk.
The donor’s email arrived Thursday morning. It was addressed to Serra, and Serra forwarded it to Mariana with a note that said, “Please coordinate.”
The donor was a retired businessman from Lyon who had funded the sequencing because he liked the idea of having a species named after him. His email asked when the press conference was. It said he had invited his daughter, who was flying in from Montreal, and he wanted to know if there would be a photographer.
Mariana wrote back that the manuscript was under review and that the naming announcement would follow publication. She did not say that the name might be wrong. She did not say that the organism might be a contaminant. She said the manuscript was under review, which was true, and she said the naming announcement would follow publication, which was also true, because the registry published names after the manuscript was accepted.
She stared at her own email for a long time after sending it.
The plasmid showed up on Friday. She had been running the genome assembly in the background while she worked on the reads, and the assembler had flagged an extrachromosomal element: a circular DNA molecule of about 4.2 kilobases, present in three copies per cell, carrying six open reading frames, none of which matched anything in the databases.
She checked the assembly. She checked the coverage. The plasmid was real. It was in the organism. It was not a chromosomal fragment misassembled into a circle, because the read depth was uniform and the ends joined cleanly.
She checked the reagent databases. The plasmid did not match any vector, any cloning backbone, any standard reagent. It was not pUC19. It was not pBR322. It was not the lab’s standard plasmid for the halophile work, which was a 3.8 kilobase vector she had built herself two years ago and which shared no sequence with this one.
A novel plasmid in a known contaminant was either a new gene or a reagent artifact. If it was a new gene, it was the most interesting thing in the genome, and it was the thing that would make the organism worth naming. If it was a reagent artifact, it was contamination from a reagent she could not identify, and the whole genome was suspect.
She went to Serra on Monday.
“There’s a plasmid,” she said. “Four point two kilobases. Six ORFs. No database match.”
He looked at the assembly report. “That’s the new gene.”
“Or a reagent artifact.”
“From what reagent?”
“Unknown.”
“Then it’s not an artifact. It’s a discovery. Mariana, this is the paper. A novel plasmid in an extremophile. The name is the least of it.”
She wanted the plasmid to be real. She wanted the organism to be new, because if it was new, then the name was honest, and the tree was sound, and the donor’s money had bought something true. She checked the plasmid three more times, with three different methods, and each time it came back as a real, circular, novel element with no database match.
She also checked the water system. She plated a sample from the tap, and the colonies were pink, and the 16S was Halobacterium salinarum at ninety-nine point one percent, and there was no plasmid. The water-system isolate did not have it. The Atacama isolate did.
This was the ambiguity. The organism was either a new species that happened to share a barcode with a contaminant, or a contaminant that had picked up a plasmid somewhere between the Atacama and the lab. The tree said new species. The barcode said contaminant. The plasmid said maybe.
The registry form did not have a box for maybe.
The registry was the International Committee on Systematics of Prokaryotes, and its validation list was published monthly. A name proposed in a paper was not valid until it appeared on the list. The deadline for the Friday list was Thursday at midnight. Once a name was on the list, it could only be withdrawn by the people who proposed it, which meant Serra, because Serra was the corresponding author.
Mariana read the rules three times. She read them in English, which was the language of the rules, and she read them in Spanish, which was the language of her mother, who had told her once that a name was a promise you made in front of everyone.
The rules said a name could be retracted. The form had a line for “reason for retraction.” It did not have a line for “we were wrong but the name is good.” It did not have a line for “the organism is probably a contaminant but the plasmid is real.” It had “reason for retraction,” and below it, “supporting evidence,” and below that, “proposed synonym.”
She filled out the form. She attached the plasmid sequence as supplementary data. She put the donor’s name in the acknowledgments, because the donor had paid for the sequencing, and that was true regardless of what the organism was.
She filed the retraction on Thursday at four in the afternoon. The system sent her a confirmation number. She printed it and put it in her bag.
Serra called her at five. “I saw the retraction.”
“Yes.”
“You filed it without telling me.”
“I filed it as the person who found the contaminant. The form allows the discoverer to request retraction.”
“The discoverer.” He paused. “You’re not the discoverer. I’m the corresponding author.”
“I’m the person who ran the sequencing. The form asks for the person who ran the sequencing.”
He was quiet for a moment. “The name survives as a synonym?”
“The name survives as a synonym. Halobacterium donantis becomes a synonym of Halobacterium salinarum. The plasmid is in the supplementary data. The donor is in the acknowledgments.”
“And the press conference?”
“There is no press conference. There is a paper, and the paper says the organism is a contaminant that carries a novel plasmid, and the name is a synonym.”
He did not say anything for a long time. Then he said, “You cost me the name.”
“I cost you a lie.”
“The name was not a lie. The tree was sound.”
“The barcode was in the raw reads.”
“The barcode was horizontal transfer.”
“Maybe. The form doesn’t have a box for maybe. It has a box for retraction, and I filled it in.”
He hung up.
The registry published the list on Friday. Mariana checked it at her desk, with a cup of tea that was still warm. Halobacterium donantis appeared, struck through, with a footnote that read: “Retracted by discoverer; synonym of Halobacterium salinarum; novel plasmid documented in supplementary data.”
The struck-through name was still readable. The line through it was thin, and the letters were dark enough that you could see what had been there. It was like a word crossed out in a letter that you could still read if you held it to the light.
She closed the browser. She went to the lab, where the pipette tip boxes were on the shelf above the bench. She took a box, the blue one, the one with the yellow sticker, and wrote on the lid in black marker. She wrote the organism’s real name, the one that had been on the plate since the beginning, the one that lived in the water system and showed up in the negative controls every quarter.
Halobacterium salinarum.
She put the box back on the shelf, next to the others, and went back to her tea.